Fatigue Life Assessment of Cut Edges in High Strength Steel

In Cooperation With Volvo Construction Equipment AB

ILONA BARMICHO

Master of Science Thesis in Lightweight Structures Dept of Aeronautical and Vehicle Engineering

Stockholm, Sweden 2015

Fatigue Life Assessment of Cut Edges in High Strength Steel

Ilona Barmicho

A Master Thesis Report written in collaboration with

Department of Aeronautical and Vehicle Engineering

Royal Institute of Technology

Stockholm, Sweden

And

Volvo Construction Equipment

Braås, Sweden

Aug 2015

1

Master of Science Thesis ISSN 1651-7660 TRITA-AVE 2015:56

Fatigue life assessment of cut edges in high strength steel

Ilona Barmicho

Approved

Examiner

Stefan Hallström

Supervisor

Zuheir Barsoum

Commissioner

Volvo Construction Equipment AB

Contact person

Bertil Jonsson

Abstract The interests in more effective and lighter structures have increased the use of high strength

steels for higher performances. Plate materials are optimized so thinner structures and higher

material strengths are reached, this leads to the cut quality might be a new issue.

In this investigation steel thickness of 6 and 16 mm with minimum yield strength from 355 to

960 MPa are fatigue tested with constant amplitude tensile loading. The specimens were cut

using waterjet and also with thermally cut methods such as plasma and oxygen. Before fatigue

testing the cut surfaces were measured and roughness Rz values were obtained.

Empirical and analytical results of the surface roughness influencing the fatigue strength for

different steel strengths are presented.

Since thermal cutting methods have been developed over the years the FAT values are higher for

those IIW are recommending.

When the quality of the cut surface can be kept high the fatigue strength will also be higher than

those recommended. This means having a cutting process that provides smooth surfaces such as

waterjet and plasma cutting the fatigue life will be longer.

2

ACKNOWLEDGMENT

I would like to extend my gratitude to my supervisors

- Zuheir Barsoum, PhD, Sr Lecturer, KTH – Lightweight Structures

- Thomas Stenberg, PhD, KTH – Lightweight Structures

- Bertil Jonsson, Expert Weld Strength, Volvo Construction Equipment AB

for their invaluable guidance, feedback and support through this master thesis.

I would also like to thank Mansoor Khurshid, PhD student at KTH Lightweight structures for his

inputs and help with the fatigue testing. A big thank to Torbjörn Narström at SSAB.

Lastly I want to thank my family for their love, support and patience.

Ilona Barmicho

Stockholm, Sep 2015

3

NOMENCLATURE

Below are the symbols and abbreviations used in this report listed.

Notations

Symbol Description

C Capacity value in SN-curve

m Slope of SN-curve

t Thickness

K Stress intensity factor

Δσ Stress range

a Crack length

f Geometry function

C0 Crack propagation constant in Paris’ law

m0 Crack propagation exponent in Paris’ law

N Number of cycles

R Stress ratio

Abbreviations

LEFM Linear Elastic Fracture Mechanics

HAZ Heat Affected Zone

FEA Finite Element Analysis

FE Finite Element

SIF Stress Intensity Factor

SN curve Wöhler curve

FAT Characteristic Fatigue Strength

4

TABLE OF CONTENTS

NOMENCLATURE 3

TABLE OF CONTENTS 4

1. INTRODUCTION 6

1.1 Background 6

1.2 Purpose 6

1.3 Delimitations 6

1.4 Method 6

2. THEORY 8

2.1 Materials and test specimen 8

2.2 Thermal cutting 10

Laser 10

Gas 11

Waterjet 11

Plasma 11

2.3 Fatigue life calculation 11

Roughness 11

LEFM – Linear Elastic Fracture Mechanics 12

Stress intensity factor (SIF) 13

S-N curve 13

3. METHOD 15

3.1 Materials and test specimens 15

3.2 Cutting 15

3.3 Surface roughness measurement 16

3.4 Fatigue test 17

3.5 Fracture Mechanics Analysis 18

4. RESULTS 20

4.1 Surface Roughness 20

4.2 Fatigue life 22

5. DISCUSSION AND CONCLUSIONS 27

5.1 Discussion 27

5.2 Conclusions 27

5

6. RECOMMENDATIONS AND FUTURE WORK 29

REFERENCES 30

APPENDIX A: DIVISION OF SPECIMENS 31

APPENDIX B: TEST PROGRAM 32

6

1. INTRODUCTION

This section contains the background of the thesis and the purpose, delimitations and used

methods.

1.1 Background Lately high strength steels have been applied for lightweight structures in order to reduce weight

and fabrication costs and to contribute to the performance. In many kind of structures there is a

need of achieving lighter and more optimized solutions. Since there are new weld treatment

methods that enhances the weld quality; plate materials may be optimized so that thinner

structures and higher material strength are reached. Consequently the quality of the cut sections

might become a new design issue. The roughness of the plate surface and the cut edges are the

main factors affecting the fatigue crack propagation, at least when large crack-like defects are at

hand. It has been seen that the fatigue strength of the base material increases by increasing the

yield strength of steel. It is important, both for internal communication in the producing

company, but also the outsourcing of cutting to suppliers, that there is a clear and concrete

regulations with scientific foundation that describes the quality of the cut edge. Describing the

quality level so that intended fatigue life targets are reached is one of the problems treated in this

thesis work.

1.2 Purpose

The purpose of this investigation is to see how cutting materials with different methods are

affected by the surface roughness and the fatigue life of the structures. The small surface defects

might have great impact on the product. What is not known is how big impact it actually has.

Specimens with geometry according to Figure 1 are in need to be developed and then fatigue

tested in order to study the distribution in fatigue life and crack initiation point to determine the

relationship between the surface roughness and the fatigue life. Linear elastic fracture

mechanics, LEFM, should be used to get results analytically and then be compared with the

experimental results. FE-simulations shall be made to determine the stress distribution of the

specimens. A similar study has already been made before by SSAB and can be found in SSAB

Plåthandbok [1] and in [2]. The reason why it is continued is because it is intended to

complement with thicker plates and also with waterjet cutting to avoid HAZ.

1.3 Delimitations

Delimitations have been summarized in a number of points, which follow:

This thesis project is part of a bigger research project, which means that only a small part

will be included within this scope.

No metallurgy studies are included in the project.

Residual stress measurement will not be included.

Only small-scaled specimens will be tested and analysed.

1.4 Method

A literary study should be performed in order to provide more background knowledge about:

7

Cutting methods and how they are performed.

Surface roughness measurements, how it is performed.

Fatigue life assessment.

Measurements

Surface conditions (e.g. roughness)

Dimensions

Steel strength

FEA – Finite Element Analysis

Fracture mechanics

Fatigue

Fatigue testing

Tensile testing

Comparison with FEA

8

2. THEORY

This section presents the theoretical reference frame that is necessary for the performed project,

such as the material, cutting methods and calculations.

2.1 Materials and test specimen

The tested specimens were manufactured from high strength steel with varied thicknesses from

16 to 40 mm with minimum yield strengths from 355 to 960 MPa. In order to distribute the stress

concentration to the middle, the specimens were shaped as dog bones with geometry according

to Figure 1 below.

Figure 1.”Dog bone” shaped specimen with geometry for fatigue test.

Due to limitations in fatigue test rig-capacity specimens should be small, there is also a possible

influence from HAZ – heat affected zone. Since the specimens were cut with different methods

that require heat it can cause severe damage to the work piece. Therefor 20 specimens were

produced with two different widths to compare the fatigue life of the two different kind of

specimen. According to Table 1 provided by SSAB the fatigue life should not be affected

considerably, which also showed when testing the specimens. The simulated geometry is

therefore acceptable.

Table 1. Heat affected zone for thermal cuttings.

t [mm] 15 30 50

Laser cut HAZ [mm] 1 - -

Gas cut HAZ [mm] 3 5 7

Plasma cut HAZ [mm] 1.5 2.5 4

The test specimens were manufactured from 6000×2500 mm steel plate, Table 2 below shows

how many of each steel grade, thickness and cutting method needed to be manufactured in order

to get proper amount of test specimens. How the specimens were cut out of each steel plate can

be found in Appendix A.

9

Table 2. Amount of specimens for testing.

t [mm] Re 355 Re 700 Re 960

Laser Gas Plasma Laser Gas Plasma Waterjet Laser Gas Plasma

6 - - - - - - 15 - - -

16 45 45 45 30 30 30 - 30 30 30

25 - 15 15 - 15 15 - - 15 15

40 - 15 30 - 15 30 - - 30 30

Around ten specimens of every thickness, steel grade and cut method had to be tested to get

relevant and comparable results. The total amount of specimens tested at KTH was 180, since

limitations with the hydraulic fatigue test machine. All of these are not covered within this

thesis; the complete results will be published elsewhere. In Table 3 below it is shown which are

included in the thesis work and Table 4 shows the chemical composition of the steel used.

Table 3. Specimens included in this thesis project.

t [mm] Gas cut Waterjet cut Plasma cut

6 - S700 - 16 S355 - S355 16 S960 - S960

Table 4. Chemical composition of the steel tested.

Grade C Si Mn P S Cr Ni Mo V Ti Cu Al N Nb B

S355 0.15 0.34 1.41 0.006 0.001 0.04 0.07 0.021 0.037 0.008 0.01 0.034 0.008 - -

S700 0.14 0.30 1.14 0.014 0.001 0.30 0.05 0.169 0.014 0.008 0.01 0.040 0.003 0.001 0.002

S960 0.17 0.23 1.26 0.008 0.001 0.20 0.06 0.600 0.041 0.003 0.01 0.052 0.003 0.015 0.001

All specimens were named in one consistent way. They were marked with the yield strength, the

cut process and sample number according to Figure 2.

Figure 2. Naming of specimens.

10

After the cutting, the edges were grinded from the plate side to ensure a sufficient connection

between the cut surface and the fatigue strength, without influence from the cut edges surface

conditions.

Waterjet cutting will also be investigated since no heat is added in the cutting process and

therefore there will be no HAZ in the material. The specimens cut by the waterjet were smaller

than the other specimens due to limitations in the cutting range in the waterjet machine. The

proportions are the same as for the other specimens, see Figure 3.

Figure 3. Geometry of specimen cut with waterjet.

2.2 Thermal cutting

Thermal cutting is most often used as preparation of the material before welding. It is a way of

cutting that has been developed considerably well, especially laser and plasma cutting [3]. In this

thesis only three different methods were used, gas, plasma and waterjet cutting. This to compare

in what extent the cut edge with its roughness is affecting the fatigue life. One significant

occurrence when cutting steel thermally is the phenomena denoted HAZ, heat affected zone as

mentioned before. It means that the material near the cut gets thermally damaged, the material is

not melted and its microstructure and properties has changed due to the heat. This is why

waterjet cutting will also be investigated since no heat is added in the cutting process and

therefore there will be no HAZ in the material. The methods are listed and explained below, for

this project the plasma and gas cutting was carried out at a supplier BE group and the waterjet

cut specimens at the mechanical department at KTH.

Laser

The laser beam with a diameter of 0.05-0.25 mm is focused by a lens or mirror against the work

piece where the beam is concentrated to a little point, which then melts, burns, vaporises or

blows away materials. During the laser cutting process an assisting gas is used to protect the

surface for oxidation reactions. The temperature gradient generated in the cutting section

becomes high during the process; which results in high residual stress levels while reducing the

end product quality. The advantages of laser are the accuracy, speed and flexibility [4]. Laser cut

edges also becomes straight, shiny and thin with small loss of material. The influence of the laser

cutting is characterized as the profile ratio of the striations formed in the cut edge surface and

this has a major influence.

11

Gas

When cutting metal using gas, most often acetylene or propane is used. The gas combusts in

oxygen and the flame with high temperature is then used to preheat the work piece. When

sufficient high temperature is reached a beam of oxygen perform the actual cutting process by

combustion of the metal.

Waterjet

Waterjet cutting uses only water, when cutting in soft and porous materials, for harder materials

sand is added in order to cut through the material. No heat exposure or thermal stresses occur

that affect the results. What is important in this case is to realise that the water speed is affecting

the surface roughness and near edge stresses [5]. The angle of the surface striations depends on

the relationship between the traverse cutting speed and water pressure. Increasing the speed at

given pressure increases the angle of striation until a through cut is no longer achieved. Also if

the nozzle is holding straight above the work piece maximum surface roughness is obtained. [5].

For this case the specimens were cut with a cutting speed of approximately 5 m/s and a pressure

of 3400 Bar.

Plasma

Plasma cutting is a process used to cut steel using plasma torch. The temperature of the plasma

arc melts the metal and pierces through the work piece while the gas flow removes the molten

material of the cut. Two important parameters that can be changed in plasma cutting are the

current intensity and cutting speed, which is what determine the quality of the cut-edge surface

and the width of the HAZ [6].

2.3 Fatigue life calculation Parent material that has not been welded could be critical for the fatigue life, where the loading

capacity is limited by the maximum stress that requires for initiating a fatigue crack.

Roughness Since the fatigue strength for parent materials is increasing with increased yield strength the

surface quality is of importance. The fatigue life of thermally cut specimens are depended of the

different cutting methods and cut qualities. This is why the roughness of the cut edges needed to

be measured. Here the results from the measurements are expressed as peak to valley height Rz

and average height of the surface roughness Ra, see Figure 4 and Figure 5. They are calculated

according to in equation(1) and equation(2).

Figure 4. Explanation of Rz.

12

5 5

,max ,min

1 1

1

5z i j

i j

R z z

(1)

Figure 5. Explanation of Ra.

0

1L

aR z dxL

(2)

LEFM – Linear Elastic Fracture Mechanics To predict crack growth in structures subjected to fatigue loads linear elastic fracture mechanics

(LEFM) is used. It is based on the assumption that there are defects in the structure and that the

stresses at the crack tip tend to infinity. This is the only method that takes the crack growth, from

the initial stage into the material, into account. A crack can be loaded in three different ways, see

Figure 6.

Figure 6. Crack modes: opening mode, in-plane shear and out-of-plane shear.

Only Mode I crack propagation will be considered within this project, since the expected crack

plane is perpendicular to the applied tensile loading. A disadvantage with LEFM is the extra

working effort needed since several different FE-simulations must be performed in order to get

enough points to describe the integration in Paris power law.

When calculating the number of cycles to failure the integration in Paris law, equation(3), is

used, where both C0 and m0 are material constants here 12

0 5 10 , /C MPa m m cycle and

0 3m with a failure probability Pf=50% according to IIW recommendations [7] the stress

intensity range ΔKI equation(4) is implemented. It depends on the length of the crack, a, stress

range, Δσ, and f which is the function of geometry loading, equation(5), with a width 18w mm

of the specimen.

13

m

I

daC K

dN (3)

IK a f (4)

0

0 00 0

0

2tan cos 0.75 2.02 0.37 1 sin

2 2 2

ma aa aw

fa w w w w

(5)

When solving the fatigue life it becomes as follows, see equation(6).

1

( )

f

i

a

m m

a

daN

C a f a

(6)

The stress intensity factor describes the stress rate near the crack tip and is calculated both

analytically and with finite element analysis FEA.

Since SIF depends on the crack length a, the integration is solved numerically using MATLAB.

Stress intensity factor (SIF) The stress intensity factor, SIF, is important to determine in linear elastic fracture mechanics and

can be determined analytically with use of the equation(7) below.

I nomK a f (7)

Where σnom is the nominal stress, a crack length and f is the function of geometry and loading.

When calculating the number of cycles to failure the formula Paris law is used and the stress

intensity range ΔKI is implemented, this is explained below.

I,m ,minI ax IK K K (8)

The SIF’s describe the stress state near the crack tip for the three modes and can be calculated

both analytically and with the use of FEA.

S-N curve In order to plot the fatigue strength of a component an S-N diagram, also named Wöhler curve is

used. The logarithm of the fatigue strength as a function of the logarithm of number of cycles in

a measure point is plotted. Usually the test results are distributed and that is why the median is

formulated which gives a failure probability of 50 %. This means that half of the samples are

expected to fail for that specific stress range and corresponding number of cycles.

For fatigue analysis of weld it is more convenient to use the stress range to describe the fatigue

life since residual stresses close to the yield limit can be expected in both tension and

compression. When compressive stresses are applied to a weld where the residual stress is near

the yield limit in tension, the weld will experience a positive stress range even though the weld is

loaded in compression. The stress range is calculated according to equation (9) below and R the

load ratio according to equation(10).

max min (9)

14

min

max

R

(10)

In order to calculate the number of cycles to failure the equation(11) below is used, where m0 in

this case is equal to 5 since it is parent materials.

0 0log log logN C m (11)

which gives the fatigue life

0

0

m

CN

(12)

and where C0 is

06

0 2 10m

C FAT (13)

it leads to fatigue life according to equation(14)

0

62 10

mFAT

N

. (14)

Where, according to IIW [7], the FAT-value gives the characteristic fatigue strength at 62 10

cycles. The characteristic fatigue strength is the stress range Δσ or σr for the mean fatigue

strength minus 2 standard deviations, 2.3 % failure probability.

15

3. METHOD

It is investigated how great impact of surface defects has on the fatigue life of steel. Steel

specimens were cut with four different methods, laser, plasma, gas and waterjet. For each of

these methods the surface roughness was measured and then fatigue tested with a fatigue-testing

machine at KTH. The stress intensity factors were obtained and the fatigue life calculated as it

can be read below.

3.1 Materials and test specimens As already has been written there were some restrictions with the hydraulic machine, only a

small part of the project were included in this thesis. Therefor only some of the total amount of

specimens were tested and investigated. In the Table 3 it can be seen which specimens were

fatigue tested and included in the thesis, they all had thickness of 16 mm except for waterjet cut

S700 steel which was 6 mm thick. The results can be found in section 4. In Figure 7 the surfaces

of the specimens are shown.

Figure 7. Explanation of the surfaces and edges of the specimen.

3.2 Cutting As written before the specimens were cut with different methods and at different places. Volvo

CE in Braås performed the laser cutting while the plasma and gas cutting at a supplier BE group

and the waterjet at the mechanical engineering department at KTH. Edge inhomogeneities were

removed using a belt grinder, see Figure 8 below.

Plate surface

Cut surface

Cut edge, in M1

Cut edge, out M2

16

Figure 8. Belt grinding the specimens.

3.3 Surface roughness measurement It is important to measure the surface roughness to be able to draw parallels between the

roughness and the fatigue life. The measuring was performed at Volvo Construction Equipment

in Eskilstuna with use of a Taylor Hobson profilometer with a two micrometre needle; see Figure

9.

Figure 9. Surface roughness measurement.

At first three positions along both cut edges where measured, see Figure 10. When noticing that

the roughness in the middle of the edge, M3, was too small compared to the sides, M1 and M2,

these results were excluded. It was considered that the outer parts near the edges would be the

most critical ones.

17

Figure 10. Where on the specimens the measurements were made.

Ra is plotted against Rz later on in section 4.1 Figure 13 and Figure 14, where Ra is the profiles

arithmetical mean deviation for the reference length and Rz the average height of the surface

roughness. In this report Rz refers to an average of five measurements. According to [8] the cut

can be classified to range 1-4 depending on what method was used, see results in section 4.1.

3.4 Fatigue test The fatigue tests were performed in a 200 kN hydraulic testing machine at KTH, Figure 11, and

in an Amsler 1000 kN pulsator at SSAB. The test frequency at KTH was 20-30 Hz and the tests

were performed in tension with constant amplitude loading with R=0.05. Stress-life fatigue tests

that achieved 65 10 cycles were considered being run out tests. The stress was limited by the

yield strength of the material. The fatigue life calculations were done according to IIW

recommendations [7]. A 2.3% probability to failure using a dual sided confidence interval for 2

million cycles was calculated using an inverse slope of m=5. The results can be seen in section

4.2.

18

Figure 11. Fatigue testing at KTH.

3.5 Fracture Mechanics Analysis To be able to compare the experimental results, analytical calculations were made. Fracture

mechanics formulas were used, such as Paris’ power law where plate thicknesses of 16 mm were

investigated. The initial defect in the integral was considered being the surface roughness values

Rz 10 – 400 μm and the life is assumed to be finished when the crack reaches half the plate

thickness. The formulas used are the same formulas presented in section 2.3.

AFGROW is a program used by the United States Department of Defence and used to validate

and predict the fatigue life of structures. This program has been used to determine the fatigue life

of an edge crack in a rectangular specimen with different thicknesses with loading in tension.

AFGROW defines an elliptical crack and together with the Walker equation(15), it confirms the

values from the FE-analysis. The thicknesses of the specimens were set the same as when

calculating with Paris’ law and with a width W = 18 mm. The crack shape ratio a/c was set to

vary from 1 to 10, where c = Rz same values as before and the tensile stress to 100 MPa, the

material constants where set to m=3 and n=0.5. In Figure 12 below it is shown how the geometry

in AFGROW is defined.

(1 )1

m

n

da KC

dN R

(15)

19

Figure 12. How the geometry is defined in AFGROW.

The results computed with AFGROW were compared to the results computed with use of

Matlab. The same results were obtained and are presented as analytical FAT-value in section 4.

20

4. RESULTS

In this chapter the results that are obtained with the methods described in the method chapter

are compiled, analysed and compared with the existing knowledge and theory presented in the

frame of reference chapter.

4.1 Surface Roughness The results from the surface roughness measurements can be seen in Figure 13 and Figure 14

below, where also the different ISO ranges for roughness are marked according to [7] for

thickness 16 mm. Figure 13 shows the roughness of the side where the cutting flames went in,

M1 in Figure 10, and Figure 14 shows the roughness of the opposite side, M2 in Figure 10.

Figure 13. Surface roughness for the tested specimens on side M1of the cut edge.

21

Figure 14. Surface roughness for the tested specimens on side M2 of the cut edge.

Table 5 shows the ranges of the roughness, the maximum and minimum Rz value for each batch

that has been measured and tested.

Table 5. Measured Rz in μm.

Type Gas cut Waterjet cut Plasma cut

S355 S960 S700 S355 S960

Rz [μm] 20 - 140 45 - 200 23 - 37 20 - 170 15 - 56

In the Figure 15-17 below the roughness distributions have been plotted in box diagrams. The

middle lines in each box shows the median value of the roughness, the upper part of the box

represent the third quartile which is the number for which 75% of the data is less than that

number and the lower part of the box represent the first quartile which is the number for which

25% of the data is less than that number. “Edge in” corresponds to M1 and “edge out” to M2 in

Figure 10. It can be noticed by looking at the charts below that M2 is always rougher than M1.

Figure 15. Roughness distribution gas cut.

0

20

40

60

80

100

120

140

160

180

200

355 t16 edge in 355 t16 edge out 960 t16 edge in 960 t16 edge out

Roughness Gas cut

22

Figure 16. Roughness distribution for waterjet cut.

Figure 17. Roughness distribution for plasma cut.

4.2 Fatigue life In this section the results from the fatigue testing and the analytical results are presented in

mainly the FAT-value.

The FAT-value at 50% means that there is a failure probability of 50% which is what it is

considered to be when testing experimentally. The evaluated fatigue classes refer to 62 10

cycles and are shown in Table 6.

0

20

40

60

80

100

120

140

160

180

200

700 t6 edge in 700 t6 edge out

Roughness Waterjet cut

0

20

40

60

80

100

120

140

160

180

200

355 t16 edge in 355 t16 edge out 960 t16 edge in 960 t16 edge out

Roughness Plasma cut

23

Table 6. FAT-values from both experimental results and analytical.

Edge condition FAT 50% FATanalytical

Gas Cut S355 298 186

Gas cut S960 204 219

Waterjet cut S700 352 276

Plasma cut S355 225 183

Plasma cut S960 162 255

The results from the fracture mechanic analysis are presented in Figure 18 below, as written in

section 3.5 the initial defects are considered being the surface roughness values Rz 10 – 400 μm.

Six different curves are plotted depending on what Rz value is chosen. The life of the specimen

is considered finished when the crack reaches half the plate thickness, which in this case is 8

mm.

Figure 18. Curve from the fracture mechanical analysis.

Figure 19 to Figure 23 represents the SN-curves for each fatigue tested batch; the red crosses are

run out results specimens that did not crack before 5 million cycles. In the all plots the blue dots

represent the failures for each tested specimen and the red crosses are the run outs. The red lines

in the plots to the left at each row represent the analytical FAT values from each batch. The red

lines in each plot to the right represent the FAT values of both the minimum Rz value and the

maximum Rz value from each batch. Specimen number, stress range and corresponding amount

of cycles can be found in Appendix B.

100

1000

10000 100000 1000000 10000000

Stre

ss [

MP

a]

Cycles N [cycles]

Rz=10 µm

Rz=50 µm

Rz=100 µm

Rz=200 µm

Rz=300 µm

Rz=400 µm

24

Figure 19. S-N curve for steel with Re 355 MPa and gas cut.

Figure 20. S-N curve for steel with Re 960 MPa and gas cut.

Figure 21. S-N curve for steel with Re 700 MPa and waterjet cut.

10

100

1000

10000 100000 1000000 10000000

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S355 Gas cut

Failures

Run outs

Natural mean curve

Mean curve (m=5)

Char. curve (FAT)

FAT 18610

100

1000

10000 100000 1000000 10000000

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S355 Gas cut

Failures

Run outs

Natural mean curve

Power (Analytical curve (Rzmin))

Power (Analytical curve (Rzmax))

10

100

1000

10000 100000 1000000 10000000

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S960 Gas cut

Failures

Run outs

Natural mean curve

Mean curve (m=5)

Char. curve (FAT)

FAT 219

10

100

1000

10000 100000 1000000 10000000

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S960 Gas cut

Failures

Run outs

Natural mean curve

Power (Analytical curve (Rzmin))

Power (Analytical curve (Rzmax))

10

100

1000

10000,0 100000,0 1000000,0 10000000,0

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S700 Waterjet cut

Failures

Run outs

Natural mean curve

Mean curve (m=5)

Char. curve (FAT)

FAT 27610

100

1000

10000 100000 1000000 10000000

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S700 Waterjet cut

Failures

Run outs

Natural mean curve

Power (Analytical curve (Rzmin))

Power (Analytical curve (Rzmax))

25

Figure 22. S-N curve for steel with Re 355 MPa and plasma cut.

Figure 23. S-N curve for steel with Re 960 MPa and plasma cut.

10

100

1000

10000 100000 1000000 10000000

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S355 Plasma cut

Failures

Run outs

Natural mean curve

Mean curve (m=5)

Char. curve (FAT)

FAT 18310

100

1000

10000 100000 1000000 10000000

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S355 Plasma cut

Failures

Run outs

Natural mean curve

Power (Analytical curve (Rzmin))

Power (Analytical curve (Rzmax))

10

100

1000

10000 100000 1000000 10000000

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S960 Plasma cut

Failures

Run outs

Natural mean curve

Mean curve (m=5)

Char. curve (FAT)

FAT 25510

100

1000

1000 10000 100000 1000000 10000000

No

min

al s

tre

ss r

ange

[M

Pa]

Cycles

S960 Plasma cut

Failures

Run outs

Natural mean curve

Power (Analytical curve (Rzmin))

Power (Analytical curve (Rzmax))

26

In Figure 24Error! Reference source not found. below the FAT-values for each batch with

median surface roughness are presented. What are also plotted are the values from SSAB’s

investigation that can be found in SSAB Plåthandbok [1] in section 5.4. The plotted values are

the FAT-values of steel with yield strengths 355, 700 and 960 MPa and at each yield strength

three different surface roughness, 30, 60 and 120 MPa. As can be seen below the SSAB results

correspond quite well with the experimental and analytical results of steel with yield strength

700 MPa. The SSAB values correspond better to the analytical results than the experimental.

Figure 24. Median surface roughness for each batch.

Rz=50

Rz=50 Rz=110

Rz=110

Rz=30

Rz=30

Rz=60

Rz=30

Rz=30

80

130

180

230

280

330

380

0 200 400 600 800 1000 1200

FAT5

0%

[M

Pa]

Yield Strength [MPa]

355 GAS experimental

355 GAS analytical

960 GAS experimental

960 GAS analytical

700 WATER experimental

700 WATER analytical

355 PLASMA experimental

355 PLASMA analytical

960 PLASMA experimental

960 PLASMA analytical

SSAB Re=355 MPa

SSAB Re=700 MPa

SSAB Re=960 MPa

Rz=30

Rz=60

Rz=120

Rz=120

Rz=60

Rz=30

Rz=120

Rz=60

Rz=30

27

5. DISCUSSION AND CONCLUSIONS

A discussion of the results and the conclusions that has been drawn during the Master of Science

thesis are presented in this chapter. The conclusions are based from the analysis.

5.1 Discussion

The purpose of this project was to investigate how the cut surfaces influence the fatigue life, this

in order to know in what extent and way it is possible to save weight.

Before measuring the surface roughness slag and other oxides from the cutting had to be

removed from the specimens. Because of the slag some surface roughness results might not have

been completely accurate; there are some uncertainties in the results. When measuring it was not

possible to get the roughness in the middle of the specimens since the needle couldn’t reach

down but this would probably not matter that much.

It was considered that this thesis would cover variable amplitude load testing but since the steel

with yield strength of 355 MPa is very ductile it was impossible to get any crack failures. Also

since the hydraulic machine only allowed dynamic load up to 200 kN. After some testing

without any failures the approach was change to constant amplitude instead. This would

probably have worked if the cut edges were rougher. It was also important to not load with

higher levels than the yield strength since the steel starts to plasticise and the properties changes.

The material behaves then differently than expected.

Since some of the specimen edges were ascut by the supplier the specimen started to crack at the

plate surface instead of the cut edge. This happened because the plate surface became rougher

than the cut edge, since it has an influence the surface of the plate should be investigated more

thoroughly. Edge preparation probably gives a bigger scatter in the roughness and it should

therefore be specified a requirement for the surface roughness of the plate surface and the edge

between the plate surface and the cut edge as well, see Figure 7.

During the thesis project some problems have occurred, such as machine breakdown and late

deliveries of specimens which has delayed the project. That is why there was no time to run

more tests. It is preferable to have more results to compare that would have given more reliable

results.

5.2 Conclusions

The surface roughness is generally rougher on the M2 side in Figure 10 and the scatter is also

larger. Increasing of the yield strength gives a rougher surface for the gas cutting method and

less rough for the plasma cutting.

Thus, if cutting thicker specimens and a high surface quality, i.e. less rough surface, is needed, it

is recommended to use Plasma cutting. The scatter in analytical and experimental fatigue

strength, see Figure 23, is probably influenced by the manual edge preparation. Further testing is

needed

Figure 19 to Figure 23 shows how well the analytical calculations correlate with the

experimental. The plasma cut steel regardless of yield strength correlate well and the gas cut

28

steel with yield strength of 355 MPa also does. It is therefore not clear if it is only the high

strength or the low strength steels that correlate better than the other.

The results show that if the quality of cutting can be kept high the fatigue strength will be higher

than those recommended in by IIW. This means that having a cutting process that provides a

smooth surface the cracks will not initiate because of the surface roughness in first hand.

Therefore waterjet cutting and plasma cutting are good alternative in order to get high fatigue life

not affected by the surface roughness of the cut edges.

29

6. RECOMMENDATIONS AND FUTURE WORK

In this chapter, recommendations on more detailed solutions and/or future work in this field are

presented.

Perform some more fatigue tests of the same batch as already been tested and with more

scattered loads, and re-measure the surface roughness of those. Since there was a lot of slag on

the surfaces there is probably in need of some more tests. Also continue with the fatigue tests of

the other batches. This in order to get more results to compare with and also more accurate once.

Be aware of where the crack starts and make sure that the cut edge has the roughest surface

otherwise it will not crack within right number of cycles.

One should also measure residual stresses in specimens of each batch to be able to set accurate

loads when testing and to be able to compare the right experimental and analytical results.

Fatigue testing with variable amplitude loading should be executed and compared with the

results from constant amplitude.

30

REFERENCES

1. SSAB Plåthandbok

2. J-O. Sperle, “Influence of parent metal strength on the fatigue strength of parent material

with machined and thermally cut edges”, IIW Document XIII-2174-07

3. K. Weman, “Svetshandboken”, Stockholm: Liber, 2007.

4. K. Mäntyjärvi, A. Väisänen, J. A. Karjalainen, “Cutting method influence on the fatigue

resistance of ultra-high-strength steel”. Int J Mater Form Vol. 2 Suppl 1:547–550. Oulu,

2009

5. Daniel J. Thomas, “Characteristics of abrasive waterjet cut-edges and the affect on

formability and fatigue performance of high strength steels”. Journal of Manufacturing

Processes 11. 97_105, Port Talbot, 2009.

6. Daniel J. Thomas, “The influence of the laser and plasma traverse cutting speed process

parameter on the cut-edge characteristics and durability of Yellow Goods vehicle

applications”. Journal of Manufacturing Processes 13, 120–132, Swansea, 2011.

7. Hobbacher A., “Fatigue design of welded joints and components”, IIW doc. XIII-1539-96,

1996.

8. SS-EN ISO 9013:2002 Thermal cutting – Classification of thermal cuts – Geometrical

product specification and quality tolerances, Swedish Standards Institute, Stockholm, 2003.

31

APPENDIX A: DIVISION OF SPECIMENS

This is how the specimens were cut out of each steel plate.

32

APPENDIX B: TEST PROGRAM These are the tested specimen with the stress range and at how many cycles they failed or did not

fail. Run out was considered at 5 million cycles.

Specimen σr N

Specimen σr N

No [MPa] [cycles]

No [MPa] [cycles]

S355-G-03 291,65 Runout

S355-P-01 337,25 317849

S355-G-04 332,5 Runout

S355-P-02 275,5 Runout

S355-G-05 332,5 629027

S355-P-09 327,75 445716

S355-G-06 337,25 Runout

S355-P-10 323 513400

S355-G-07 327,75 Runout

S355-P-03 313,5 431958

S355-G-08 351,5 1872394

S355-P-06 285 455021

S355-G-09 365,75 290921

S355-P-07 304 1955000

S355-G-10 380 106698

S355-P-20 256,5 Runout

S355-G-11 356,25 59642

S355-P-21 332,5 355822

S355-P-22 285 702918

W960-G-02 608 100000 W960-G-03 380 414933

W960-P-04 654,55 86281

W960-G-04 475 167772

W960-P-05 617,5 86047

W960-G-05 427,5 302042

W960-P-07 408,5 231761

W960-G-06 332,5 483041

W960-P-02 570 110000

W960-G-07 522,5 151605

W960-P-03 475 100545

W960-G-08 446,5 259607

W960-P-06 437 201651

W960-P-08 475 214592

W700-W-01 665 147197

W960-P-10 408,5 246113

W700-W-02 380 Runout

W960-P-01 437 121498

W700-W-03 475 3113018

W960-P-11 475 922638

W700-W-04 570 352957

W960-P-12 475 247969

W700-W-05 522,5 345066 W700-W-06 617,5 102296 W700-W-07 641,25 178079 W700-W-08 617,5 144254 W700-W-09 522,5 Runout W700-W-10 593,75 171000