Summary of fatigue investigations for shipbuilding

(1)

SUMMARY OF FATIGUE INVESTIGATIONS FOR SHIPBUILDING

by

J.J.W. Nibbering

Program sponsored by the European Community and the Ship Research Institutes of Belgium, France, Germany, Italy and the Netherlands.

Ship Structures Laboratory, March 1975

Delft University. of Technology, Mekeiweg 2 - DELFT,

The Netherlands. Report No.

(2)

-2-Introduction.

For shipbuilding it is ofgreat interest to.dispose of reliable information about the relative fatigue-strengths of structures made of Steel 42 and Steel 52..

The existing literature does not met this need satisfactorily. 'Specimens and loads are often not sufficiently representative for ship structures Also the criteria applied may differ from what is relevant. in ships cracks always start at weld defects at structural discontinuities. They may propagate to large. lengths, befOre becoming danger.òus. This makes that generally information about the propagation stage of crack development is of more interest than in-formation about numbers of cycles up to crack initiation.

Furthermore as even long cracks do not reduce the cross section of a ship substantially, they have little influence on the magnitude of the nominal stresses. Consequently results of constant-load tests will be too pessimistic for ships.. Apart from that it is obvious that the loading of ships is not con-stant at all. This applies both to the amplitude of the loads as to the mean values.

Discussions between representatives of the Ship Research Centres and Laborat-ries of 5 European countLaborat-ries led tó the conviction, that primarily investiga-tions with full size structural models were required in order to obtain direct indications about the fatiguestrengths of ship structures made of St. Lt2 and

St. 52. However both fabrication and testing of theseS is expensive and time-consuming. Moreover only few laboratories are able to carry out such

experim-ents.

Therefore next to a limited number of complex structures smaller and simpler structural models were designed' to be tested at 100 tons-fatiguetesting machines. Finally unwelded notched plate specimens completed the range of test specimens. The three types mentioned form altogether a model family. .They were provided with a number of strain gauges in order to be able to

study correlation's between the experimental results on the basis of signific-ant (local) stresses. . . . .

The corre Zations found are sufficiently accurate to allow the conclusion that predictions of the fatigue-performance

of

complicated structures are possible on the basis

of

results

of

experiments with simple models.

This increases greatly the value of the results of the notched plate specimens.

-The...greater _par.tof_the ines:tigatiq has, been, reported in / 1/..

The Belgian experiments with continuous longitudinals (no interrup&t

transverse bulkhead subjected to bending) have been analyzed by ir. I-I.G.. Scholte and reported in /2/. '

The corresponding Italian experiments as originally conceived in the test pro-gram are also ready. The results were not satisfactory partly due to the small number of specimens 'and to differences between weld-details in the specimens.

The italian delegation has promised to let fabricate and test a number pf new specimens for their own account. The testing has already started 'and the re-suits will be made available to the European Community aS soon as possible.

Discussion of progräm and results. '

Figures 2-6 of the appendix show the specimens tested.

A summary of the conclusions from the experiments together with some final observations is given below. ' '

-a. German experiments: (crack-growth data not available).

1. Inthe region of ±'elatively low.numbers of cycles (i0-l0) there is a clear advantage for' St. 52'over St. '42. This has 'been' observed for both'

(3)

A large influence of mean stress was found. At IO5 cycles alternating load-Ing between -1-168 N/mm2 and -168 N/mm2 was aequivalent to repeated loading with a maximum stress value of 195 N/mm2.

The importance of mean stress was confirmed in the experiments with. program loading.

A simple 3-stage .program with large variations of mean stress resulted in markedly superior behaviour of St. 52 over St. '42.

A multistage program with constant mean stress resulted in a much smaller advantage for St. 52.

b.

Belgi w-z experiments.

1, Conclusion a.1 of the German experiments is only slightly supported by the Belgian data. In fact only for axial alternating loading and bending re-peated loading some advantage for St. 52 was found.

It is of interest that the experiments for axial and bending repeated. load-ing lead to the conclusion that .the larger the crack lengths the more the results converge. This tendency was not observed for alternating loading..

(Figs. .11 and 12).

2 Conclusion a.2 of the German experiments applies generally also to the Bel-.gian data, but it became clear that mean stresses had mainly effect during

the stage of crack propagation. For the stages of crack initiation no sig-nificant influence could be observed. (Figs. 11 and 12).

3. Between the values c and m in the formula .

da m

c(1XK)

a unique relation was found for axial and bending.repeated loading:

-5 -'4,22 m

c

5,710

e

..

It applied as well to St. '42 as to St 52 and to any of the three thicknes-ses 13, 19 and 25 mm.

Alternating loading did not show the same relation. This may be attributed to closing of the cracks during the compressive part of the load cycles. This closing depends strongly on yield strength.

4 The fatigue-bending experiments with large size structural models resulted in a clear advantage for St. 52 over St. '42. (Fig. 8).

-

.c.French .experiments..

The results of the experiments- with structural models in France,, contrast sharp-ly with those of the Belgian and German experiments with notched plates.

No advantage for St52 over St. 42 for both P . ./ 1 and -min max

(Fg,. .23). .

No influence of mean stress, not even for the larger crack lengths! The main explanation for the great differences in behaviour between welded structural and unwelded notched specimens may be that In the. former residual stresses are present, which are higher the higher the -yield point.

d.

Dutch experiments.

Although not large, a distinct advantage for St. 52 appeared (io 15% at.. 1O'4 cycles), which became more pronounced the larger the crack length.

It should be reminded that an advantage for St.-5.2 was not observed with the French experiments with small structural models.. (Fig. 38).

The influence of test temperature was important.

e. Italian

experiments.

(See- also introduction).

Comparing t.he results of the 4 specimens of St, 52 subjected to 3-step loading with the Belgiarrexperiments for constant loading, no contradiction was found

(4)

f.

Final observations.

It is remarkable that some conclusions which have been drawn from the results of the simply notched specimens differ essentially from those for the welded structural specimens. Due to that, a straight-forward answer to the question whether higher strength steel behaves better or not than mild s1eel under,

cyc-lic loading is difficUlt to give.

The results of the Germanexperiments with non-welded specimens show clear ad-vantages for st. 52 over St. 42 in the region of higher stress amplitudes. This

is not confirmed by the Belgian results. The large structures tèsted in Deift, Belgium and Italy support this tendency. Furthermore the non-constant load tests with widely differing programs (notched specimens) showed a superior behaviour

for St. 52. However, the French results are in opposition. Taking all results together, the advantage for St. 52 is apparent. But obviously unknown factors may cause. deviating results. A closer analysis of results, fracture areas, and physical properties of the 'fractured material, together with some additional

testing may clarify the picture in the.future.

An important part of the whole investigation has been the search for correla-tions between the results of structural and non-structural specimens.. Many small strain gauges have been used in order to be able to present the data in tçrms of "nominal" or "local" stresses. A satisfactory correspondance was foundbetwee.n the results in terms of local stresses for the large and complicated structures tested in Deift and Belgium, and' the smaller simplified structures tested in France.

Rather satisfactory correspondance was found between the results of on one hand structural and on the other hand notched plate. specimens. For instance in figs. 47 and 49 the curves 2 .for local stresses in the structural specimens (P .

I

min P -) fall between the curves P . IP O and -1 of the notched plate

max min max

specimens. '

A fracture-mechanics approach in the evaluation of the results proved to be satisfactory.

When a choice must be made between the results for P . IP O and P . IP

min max min max

-1, it seems wise to take for shipbuilding P . IP O. An additional ad-vantage is,, that it is also the cheaper type. min max

0f great importance is the -observed inf-iuence of mean stress on the. fatigue re-suits. This' was the outcome both from the experiments with constant load am-plitude and from the programmed load tests. In the latter especially the influ-ence of the variations of the mean stress level became evident. It appeared that. mild steel suffered much more from the influence of the variations of the mean stress level than higher strength steel.

References.

/1/ Nibbering, J.JW., H,.G. Scholte and J. van Lint,. Synthesis of cooperative fatigue investigations with notched plates and welded ship structures of St. 42 and St., 52. Report NSS-TÑO no. 206 5, Dec. 1914.,

/2/ Schoite, H.G., Fatigue. experiments with full-scale bottomlongitudinals of St. 42 and St. 52 under alternating bending. Report no. 195 S.S..L. Deift, March 1975.

/3/ Paetzold, H., Vergleichende Untersuchungen an gekerbten Proben aus Schiff-baustahi A Und DH 36. Bericht 33/73, F.D..S. Hamburg.

/4/ .Report 153/4585 of the "Lab. voor Weerstand van Materialen en. Lastechniek",

Rijksuniv. Ghent. . .

Report 153/4706 àf the "Lab. voor Weerstand van Materialen en Lastechniek",

(5)

-;5-/5/ Compte rendu d'essais de fatigue en traction onduie et

traction-compres-sion d'lments souds. L'Armement Lopard. Paris 1973. Ministre d'tat

(6)

(7)

Fig. 2.

Large scale specimens (Dutch experiments).

Fig. 3.

Large scale specimens (Belgian and Italian experiments). TVPEA es.«.: 1:10 1350 t' 2150

j

Ab 4 13 TYPE. B

Fig. 4. Specimens type a and b (French eperiments).

T

-

r.

j j

a..HZj_..a0 O nl tj

_._i_EfL.J._.._._

B O (550)0-joint a backgroone Is lo be chippedand rsealing n .10 run to bi made. B W.ldin como.ntions

Alt connections. are to be continuously

molded with AbS (the connection of email

flange to traneneron bracket kow.o.r jeto

be continuously moLded w,th,4h4).

Platee el equal thicko out of enechar0.. Tolerance of plate thickness: ±2mm.

-(loi .;a.a.. Tetatle«th of frame: 4000 teso through R. 10 A.A O.toil A scale 2:1, 350 POIOiIA 700 13 T bicI,: 13

('I '

X17

100

Specimen to, testing

under bending

Loading.-350

35°

Fig. 5. Specimens (German experiments)..

Speconen fer teting:under ao(e toudiog.

Fig. 6. Specimens (Belgian experiments).

250 IO mm. 350 13 9=30 -j'3 1 I- Ji

2Jotoo:2!9

- 200 -340 -050 113 c-c 500 Notd« detail., - 500 300 Tfdnk:19 j -13.J

t.

. . X ,(55)X-Joiol 13

H-500 thlcjonese::13 100.

-050 thi.n.as: 13

(8)

2 N Su.,n,ini 1,23 St 51. S36'.5 's

'

'-'K' 'y-°

_{\ \}

/\

rsp.cim.s 1: risulti CarrsCt.d for

pr.o.dn9 10Gdn2 al

Cia -Cous,

300 N /rors

b i O 9io' 2 3 4 5 6 7 B 95 2 3 4 5 6 7 _9io6 2 3 4 S 6 78

97

P16.9 RESULTS OF GERMAN EXPERIMENTS WITH CENTRALLY NOTCHED PLATES UNOER ASIIAL LOADING.

40QP SQ. 300 Prom

II

'T

. 90'%"ì"\\10% 50% /.Probability of survivt i, ¿ i

1

p -I s0

ÌI1

i_Q_ b 100 i _____ I I . . I t . i I Numbir of cyctes...,.

I,

I . 400 3-450 - 6..Q- --S S-- S-S-- S-S-- _N ---S--- S' -W-- _'s 2 -S-S-z-S 5 +

-'

N '-."

Number of cycles 250 10' 2 3 4

FIO.8 WÖHLER CURVES FOR VARIOUS CRACK AREAS 0F BELGIAN LARGE SCALE SPECIMENS; St.52,grade DH3SNb. AND St.42,grade D. BENDING LOADING mn./Tmac - 1/2

'S 5 6 7 8 9 _io 2 600 550 500 450 o -u C rs w 400

I

350

(9)

C .5 2 C C 300 250

i20

_z

i

E

Ei

700

Specimen for testing under aiat Loading.

P 75

\o

" "s Specimen for 'J testing under s.',..- s bending Loading - Notch detaiL... 400 10 450 40

r

'0 o 350 C 5 o 30 o u n o 250k

.5

E

20-j

D = St.42,grade.D DH St.52,grade. DH36Nb. thickness 13,19 or 25mm O4'

Axial repeated Loading

----Axial alternated loading

Bending repeated loading

2 3 4 5 5 7 g g 2 3 4 5 6 69

Number of cycles

-FIOEII WOHLER CURVES FOR 1mm CRACKIIENGTH OF LARGEST CRACK (BELGIAN SPECIMENS)

D St.42,gradeD OH St.52,grade.DH36Nb.

thickness 13,19 or 25mm

AeiL repeated lOading

-- Acial alternated Loading

Bending r.p.at.d loading

Number of cyclic 100 i i i

ti

2 3 4

5 6lB94

2 4 56 7091o8 I I I

Jill

3 4 5 6 7 8 9 2 3 4 5 6 7 8.9

(10)

300 200 loo o,- -300 201 100 E 300- 20G-o 17E 15)-

1004-J

50 E E z -1f. E -- mIh i t t t i U 102 2 3. 4 5 6 78 C.n orn Strain gaa..IO .Stratn ata.20 Number of cycles

-, gnom. French specirnens;Aype. St.42,grade.O (---) St52,grade. DN36Nb. ( flOm Crack Number of cyckes - -0_ 0 -i i t t I t t t i t t- I t- t I t t-t- I 5 6 7 8 ₁₀ 2 -- 3 4 5 6- 7 89 5- . 2 3 4 5 6 8 FIG.23 WÖHLERCURVES FOR-CRACK AREAS OF 0,100 and 500mm3 Atype SPECIMENS-; min./m5x _1/2

-- .

--F1038 COMPARISON BETWEEN THE RESULTS -FOR StL2 grade D AND St.52 grade DH36N BRAcXETS+ 20CC

500 mm2

- 100mm2 tSt.52,grade. DH36Nb

Omm?)

- -_iP.r11_t._ _{St42(EarLu er tests)}

Omm2

(11)

(\%

_:.

crk

N\

_-.

_i1:k.

\ im:

i

NQ

.

I'.

.'.

ior

.. -...j'

Ontek Bal)

'.'nnton

---_

---

---®

Fr.nch Ipinmim A-typa 3 2 3

o

i.2..-1 Franck .p.Ñran.; B-typi.

Valack ar., '.9 2 E z t?E E 20-z Number of cycles

-.

4 Frmuh A-typa lo I I i I

Ilt'tI

I i

111111

2 3 4 5 6 7 8 9 y5 2 3 1 , S 6 7 B 96

Fig. 47-49

Comparison of ait, results, now including

the large scale specimens tested 'in

Belgium for 5mrn cracklength (a or seq)

-I

5t42,

Dutch-, French-' and Belgian structurai

specimens: Pmjn/Pax' 1/2

St.52

Belgian notched bend' specimens:

Pmjn/max

O resp. i

o

BclO,n. cp.circ.n.

t)

-o

n,, c-a a, SS

<N

'a c-.

"t) " 'a"a

a-. c-

-1.2..1 Franck Ipacim.n,; 9-typ..

2-

curkar.rT

frZ3

_{fl1_I}

I VjUI

I

Wrack aro' Dutch sp.ómarIs(tIt (----) Vcr ach: arO' GEe Number of cycto O I I I

II-a-III

I -I

i'iiiitl

ia3 - 2 3 4

5 '6 7 B94---4 5

_{6 7 89 o5} - 2 3 1 1 t I I I-4 5 6 7 B!9i6

-.

700 crack Cr.. 19

o

I 600- % .a.Iarg.fl 600 70 Crack arta

o

;l'jlan cp.cinr.a 19 ______ - ______