Technical stell research

8/22/2019 Technical stell research

1/141

Xkc"fk"Ecuvgn"Tqocpq"322."K/2234:"Tqog


2/141


3/141

FINAL SUMMARY

High grade strength plate steels were used to improve their applications in EU obtain advantages to the

overall stiffness of the decks as well as their fatigue resistance. The yield strengths of the steels were in

the range 4601100 MPa.

The application of these products were focalised to design and to produce conventional and sandwich

panels in ship as well as for bridge application.

In order to met the objectives the project were articulated in the follow tasks:

- To define the procedures to design, by geometrical parametric inputs, conventional and sandwichdecks.

- To investigate the fatigue behaviour of the shape of welding joints, by specimens, and to comparethe results to Eurocode 3.

- To identify the methods for manufacturing heavy sandwich panels and the related weldingprocedure for laser joints.

- To manufacture conventional panels directly in shipyards in order to obtain industrial products asreference.

- To manufacture sandwich panel by laser welded joints. Two grades of steel plates (460 and 1100MPa) were selected.

- To test the conventional and sandwich panels in order to compare with EC3 and to verify theimprovement of their reduction of weight, the increasing both of the stiffness and the fatigue

behaviour

Materials

In order to manufacture conventional and sandwich decks the selection of plates were done in term

both of thickness and of grade.

The plates was delivered by Fincantier (Italian shipyards) and produced by ThyssenKrupp Nirosta

GmbH (Germany) and from Corus (UK)

Steel Grade YS

[MPa]

Thickness [mm] Supplier

355 8

560 16

690 8

690 12690 16

690 30

1100 8

355 2

FINCANTIERI

(TKS)

355 8

460 8

690 8

355 2

CORUS

3


4/141

Design

Conventional decks, with upper skin of th=16mm, and grades S560 and S690 MPa were designed by

FEA. The parameters used to optimise the weight and the stiffness was: bulb area (web thickness and

height) and web numbers. The minimum deflection obtained applying a medium load was: thickness

= 10mm, height=275mm, number of web number N=3. A number of 11 conventional decks (in

particular 5 were in grade S560 and 6 in grade 690 and reinforced bulb in grade S355) were

manufactured ready to submit under fatigue tests. The dimension of each the conventional panel was:

length L= 3000mm, width W=1500 mm, height H= 275mm, weight P=800kgm.

Sandwich decks were designed by 2D. After the valuation of the results the design was improved by a

3D FEM.

The objective of the model was to minimize the buckling of the

corrugate elements as well as the deflection of the panel. The

parameters used to obtain the expected results were: the shape and

the size of the corrugate element in term of: thickness, web/plate

angle, wave length and height.

Optimised dimensions of the sandwich decks were obtained also

taking into account the manufacturing problems to weld, by laser

plant, the internal corrugate elements to upper and lower skin

plates.

The dimensions of the corrugate internal parts are reported in the sketch on the right, where the

thickness of the sheet S355was th=2mm and the upper plates (S1100) and lower plates (S355) was

th=8mm.

Welded Joints

All the shapes (figures) of the welded joints useful to manufacture conventional and sandwich decks

were manufactured in order to investigate their fatigue resistance and compared with the requirements

of Eurocode 3.

For conventional decks MAG, SAW and SMAW welding procedure were considered as well as butt

and T-Butt joint. Welded joint plates of dimension 1000x1000mm were manufactured directly in

shipyard in order to test true industrial production.

Specimens of with W100mm and length L500mm were machined from welded plates and submitted

under fatigue tests. Fatigue tests were carried out in constant and variable amplitude in order to obtain

both the Whler and Ganer curves.

Moreover the notch effect of the welded line was evaluated as well as the real damage. All test data

clearly are on the safe side. Compared to the experimental data derived from the tests, Eurocode3

seams to be conservative.

For sandwich decks laser welding procedure was defined and two shape of spare welded joints

specimens were investigate as follows::

Laser welded joints were cut from welded plates manufactured in four series with different grade of

upper and lower plates:

TypeA:RD1100/RD355 (Series 2)

TypeA:RD355/RD355 (Series 3) TypeB:RD355/RD355 (Series 4) TypeB:RD1100/RD355 (Series 1)

4


5/141

First Material specification describes thick plate material plate material (th=8mm), second material

description describes corrugate thin sheet material (th=2mm). Type B specimens are specimens

welded from the thick plate side to the thin sheet. For Type A specimens the thin sheet is welded to

the plate from the thin sheet side.

These specimens were submitted under axial loading with constant and variable amplitude in order to

derive Whler and Ganer curves as well as the real damage.

The results lead to define appropriate FAT classes for laser welded specimens Type A and Type B,

as these details are not listed in Eurocode tables.

Possible classification for Type A was:

FAT 36 : Transverse butt weld, welded from one side without backing, full penetration (t region, thus crack initiation sites are decided by the localnotch factor value only. This behavior presumes a very severe residual stress state in the weld zone.

The unusual failure behaviour at higher load levels could have different explanations, the easiest onebeing related to the uncertainties in notch factor evaluation: specimens E13, E15, E17and E21 could

without difficulty have different geometrical features with respect to the others (it has to be

remembered that only two E specimens have been measured, and both have not been tested).

The distortion of E series specimens could also play a role. As pointed out above, the E series

specimens are quite distorted around the weld axis (up to 5) due to the welding procedure; since the

stress flux lines inside the specimen tend to follow the shortest path, evidently weld notches on one

side are less stressed (so on the other side theyre more stressed) with respect to the no distortion

condition.

The influence of the curvature on the notch factors has been investigated through a plane FEM model

reproducing the butt weld cross section; r/20 has been adopted as minimum element size in the highly

stressed regions. The actual test configuration of the E specimens and a curvature equal to 5 have been

taken into account: in this conditions the top-right and the bottom right notch factors are respectively

+5% higher and -10% lower then the ideal situation, thus such notch factors are closer if the actual

89


90/141

shape of the specimens is considered. Figure 93 shows the FEM results for two equal notches in terms

equivalent stresses of acc.to von Mises (in plane stress conditions such a stress is equal to the first

principal stress in the tensile area, and to the third principal stress in the compressive area): the top and

bottom highly stressed areas evidently experience different stress levels (the reference value of the

notch factor is 3.20).

Since the weld geometry is measured in the actual condition, the distortion is already taken into

account, thus the curvature effect can be perhaps better considered as a fictitious increment and/ordecrement of the nominal stress.

In Figure 94 the SN curves are shown in terms of local stress amplitude, that is obtained multiplying

the nominal stress amplitude for the stress concentration factor related to the initiation notch. The three

original SN curves in terms of nominal stresses can be evidently transformed to one common curve (the

local fatigue strength ranges between 278 MPa and 287 MPa for R=0). In the knee point area the

scatter is high, mainly due to the SMAW series, which is scattered in itself.

90


91/141

Figure

84-Notchradiusstatisticaldistr

ibutionfor

MAGSeries.

Figure85-Notchanglestatisticaldistributionfo

r

MAGSeries.

Figure86-Notchradiusstatisticaldistributionfor

SAWSe

ries.

91


92/141

Figure8

7-Notchanglestatisticaldistribu

tionforSAW

Series.

Figure88-Notchradiusstatisticaldistributionf

or

SMAWSeries.

Figure89-Notchanglest

atisticaldistributionfor

SMAWSeries.

92


93/141

Figure90-Notchfactorstatisticaldistributionfor

MAGSeries.

Figure91-Notchfactorstatisticaldistributionfor

SAWSeries.

Figure92-Notchfactorst

atisticaldistributionfor

SMAWSeries.

93


94/141

Figure 93 - Local equivalent stress acc. to Von Mises calculated by FEM.

94


95/141

10

4

2

4

6

810

5

2

4

6

810

6

2

4

6

810

7

2

100

100

200

300

400

500

600

700

800

900

1000

1000

MAG:

Gasmetalarcwelding

SAW

:

Submergedarcwelding

SMAW

:Shieldedmetalarcwelding

287

_

_

P

=50%

T=

1:1.35

Specim

en

TypeC

Fig16_Fat_test_local_280186.

OPJ

Fatigue

Strength

of

Buttw

elds

Fig.8

LBF280186

MPa

k=

3.5

Localstressamplitudea,n,a,n

CyclestoruptureNr,Nr

Material:

S690QL

Geometry:

Buttweld

Thickness:

30mm

Load:

Bending

Ratio:

R

=0

Constantamplitude

CA

.MA

G

MAG

CA

.SAW

SAW

CA

.SMAW

SMAW

Figure94-Fatiguestrengthofbu

ttweldsintermsoflocalstressamplitude.

95


96/141

respectively the actual endurance and the endurance expected Test results and Eurocode3: comparison

The real damage sum has been calculated for butt welds subjected to variable amplitude loading tests.

Such calculation has been performed according to the procedure illustrated in Figure 95. Partial

damages are evaluated starting from the relevant Whler curve, and summed up to cover the actual

operating life of the specimen.

Values of Dreal are reported in Table 29: The fact that the real damage is significantly higher then 1

has strongly to be noted.

Such a result is expressed differently in Figure 96 where the Ganer curves (Pf=50%) are calculated in

order to obtain D=1; for damage calculations Whler curves slope after the knee point is assessed by

Haibachs recommendation for welded steel:

1k2k =' (14)Evidently the specimens endure longer then expected making reference to a damage sum equal to one,

even if E series is quite close to the relevant Ganer curve.

In Figure 97 test results and curves suggested by Eurocode3 (latest version available: April 1992) are

shown and compared.

Eurocode3 design curves are ideally divided into three segments according to fatigue life ranges: for

normally loaded welds, up to 5 million cycles the slope is k=3, between 5 million cycles and 100

million cycles the slope becomes k=5, after 100 million cycles the inclination becomes null (cut offlimit). Every curve is so completely determined if the FAT Class is known: the FAT Class corresponds

to the nominal stress range at 2 million cycles, and its value in MPa depends on the weld geometry and

loading. Design curves refer to 2,3% of probability of rupture, i.e. 2 standard deviations less then the

mean value; it has also to be noted that Eurocode3 expresses Whler curve in terms of nominal stress

range , and not nominal stress amplitude.

Butt welds pertain to the FAT90 Class, so fatigue strength in terms of nominal stress amplitude at 2

million cycles is 45 MPa. In Figure 97 fatigue strength of Eurocode3 design curve is not exactly 45

MPa, but 43 MPa, since Eurocode3 suggests the following multiplying factor in order to take into

account the thickness effect:

25025

,

tRjR.tj

=

(15)

Whler curves obtained by tests refer to 2.3% probability of rupture in order to perform the

comparison. Clearly the SN curve suggested by Eurocode3 is well suitable to be used for safe design of

butt welds in consideration.

Figure 97 shows the Ganer curves for the three specimens series and for Eurocode3 as well: each one

is evaluated for D=1, and refers to 2.3% of probability of rupture. Derived slopes are k=3.6 and k=3

respectively.

The results illustrates above are condensed in Figure 98, 99 and 100: for each diagram the abscissa and

the ordinate arefor a damage sum equal to 1.

If D=1 is the underlying design condition, Eurocode3 provides very conservative results.

96


97/141

Modificationo

ftheS-N-Curveand

CalculationofFatig

ueLife

(schematical)

DIA4784e

i=1:steel,

aluminium

i=2:cast

andsintered

materials

exper.

s

.

spec

real

real

.

spec

s

calc.

N

LD

D

D

DL

N

=

=

.

calc

N

k'=2k-i

Cumulativefre

quency

distribution(sp

ectrum)

Woehlercurve

Gassnercurve

slope k

Dam

agesumofthespectrum:

n1

n2

n3

n4

Ls

N k

.

Spec

n1

i

ii

D

Nn

=

=

1

2

3

4

N1

N2

N3

N4

max

,a

Stressamplitudea,a

k(kneepoint)

Cy

clesN,N

k'=k

Figure95-Procedureforassessingtherealdamagesum

97


98/141

Nominal

stress

amplitude

(ideal)

Cycles to

rupture

Real

Damage

Sum

Spec.No.

SheetNo.

Test

#)

Stressratio

[MPa] Nr, r

F4 F1-F9 VA 0 360 947500 3,33

F7 F1-F9 VA 0 360 808000 2,84

F3 F1-F9 VA 0 300 1704000 3,14

F6 F1-F9 VA 0 300 1283000 2,36

F2 F1-F9 VA 0 250 4617200 4,42

F5 F1-F9 VA 0 250 2746000 2,63

AS5 AS1-AS12 VA 0 520 633054 4,47

AS7 AS1-AS12 VA 0 520 274539 1,94

AS2 AS1-AS12 VA 0 430 1255000 4,54

AS8 AS1-AS12 VA 0 430 734653 2,66

AS4 AS1-AS12 VA 0 360 5330075 10,29

AS6 AS1-AS12 VA 0 360 1587005 3,06

E5 E1-E9 VA 0 360 1061000 1,41

E1 E1-E9 VA 0 360 1093000 1,45

E6 E1-E9 VA 0 300 1700000 1,16

E3 E1-E9 VA 0 300 2505000 1,71E4 E1-E9 VA 0 250 3212000 1,09

E7 E1-E9 VA 0 250 3232000 1,10

Table 29 - Real damage sums for butt welds.

98


99/141

10

4

2

4

6

810

5

2

4

6

810

6

2

4

6

810

7

2

60

80

100

100

200

300

400

500

600

k=3.7

k=6

k=3.8

k=3.7

100 9

0

_

_

P

=50%

T=1:1.35

SpecimenTypeC

Fig18

_D

_calc

_butt

_280186.

OPJ

runo

ut(withoutfailure)

retes

tedatahigherlevel(broken)

GassnerCurves(D

=1)andVariableAmplitudeLoadingTe

stResults:compar

ison

Fig.xxx

LBF

280186

76

MPa

k=3.5

Nominalstressamplitudea,n,a,n

Cyclestorupture

N r,N r

Material:

S690QL

Geometry:

Buttweld

Thickness:

30mm

Load:

Bending

Ratio:

R=0

TestResults(ConstantAmplitude

)

CA.MAG

CA.SAW

CA.SMAW

TestResults(Gauss,L

s,G

=5104 )

VA.MAG

VA.SAW

VA.SMAW

WhlerandGassner(D=1)Curve

s

MAG

SAW

SMAW

Figure96-GanercurvescalculatedforD=1.

99


100/141

10

4

2

4

6

810

5

2

4

6

810

6

2

4

6

810

7

2

2020

40

60

80

100

100

200

300

400

500

600

k=3,6

k=

6

k=3,5

k=3

436

0

k

=5

Material:

S690QL

Geometry:

Buttweld

Thickness:

30mm

Load:

Bending

Ratio:

R=0

ConstantAmplitudeand

VariableAmplitude(Gauss,Ls

,G

=510

4)

LoadingTests:Results

CA.MAG

V

A.MAG

CA.SAW

V

A.SAW

CA.SMAW

V

A.SMAW

WhlerandGaner(D=1)Cur

ves

MAG

SAW

SMAW

Eurocode3

79

71

_

_

P

=2

,3%

T=1:1.35

SpecimenTypeC

Fig19_D_calc_butt_280186.OPJ

runo

ut(withoutfailure)

retes

tedatahigherlevel(broken)

Eurocode3(D=1)andVariableAmplitudeLoadingTestResu

lts:comparison

Fig.xx

LBF280186

32

MPa

k=

3

Nominalstressamplitudea,n,a,n

CyclestoruptureNr

,Nr

Figure97-GanercurvescalculatedforD=1.

100


101/141

10

4

105

10

6

10

7

10

4

10

5

10

6

10

7

Dreal1

D=

1,Pf=

50%

MAG

CyclesexpectedforD=1

10

4

10

5

10

6

10

7

10

4

10

5

10

6

10

7

Dreal1

D=

1,Pf=

50%

SAW

Cyclestorupture

10

4

10

5

106

10

7

10

4

10

5

10

6

10

7

D=

1,Pf=

50% D

real>1

Dreal1

Dreal1

D=

1,Pf=

2,3%

SAW

Cyc

lestorupture

10

4

10

5

10

6

10

7

10

4

10

5

10

6

10

7

Dreal1

Dreal1

Dreal1

D=

1,Pf=

2,3%

SAW

C

yclestorupture

10

4

10

5

10

6

10

7

10

4

10

5

10

6

10

7

Dreal1

Dreal

Technical stell research

Documents