basic research of weld process and microstructure modeling for … · 2019-12-16 · 1 basic...

27
1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel Final Report Submitted by Katie Strader, Bin Wang, Prof. Cuixin Chen, Prof. Wei Zhang and Prof. John C. Lippold Welding Engineering Program Dept. of Materials Science and Engineering The Ohio State University To (Version 1 - 09-21-2015)

Upload: others

Post on 20-Mar-2020

6 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

1

Basic Research of Weld Process

and Microstructure Modeling for a

Hot-Rolled High Strength Steel

Final Report

Submitted by

Katie Strader, Bin Wang, Prof. Cuixin Chen,

Prof. Wei Zhang and Prof. John C. Lippold

Welding Engineering Program

Dept. of Materials Science and Engineering

The Ohio State University

To

(Version 1 - 09-21-2015)

Page 2: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

2

Contents Executive Summary ........................................................................................................................ 3

1. Introduction ................................................................................................................................. 4

2. Research Objectives .................................................................................................................... 4

3. Experimental Approaches ........................................................................................................... 6

3.1. Gleeble-based physical simulation of CGHAZ microstructure ........................................... 6

3.2. Samples for Charpy impact testing ...................................................................................... 7

3.3. Charpy impact testing .......................................................................................................... 7

3.4. Microstructure characterization ........................................................................................... 8

4. Approaches for Integrated Weld Modeling ................................................................................ 8

4.1. Equation for grain growth in CGHAZ ................................................................................. 8

4.2. Equation for CGHAZ hardness ............................................................................................ 9

4.3. Equation for Charpy impact toughness of CGHAZ ........................................................... 10

4.4. Welding heat transfer model .............................................................................................. 10

4.5. Integrated weld modeling .................................................................................................. 11

5. Results and Discussion ............................................................................................................. 12

5.1. Welding thermal simulation in Gleeble ............................................................................. 12

5.2. Base metal microstructure.................................................................................................. 13

5.3. Gleeble-simulated CGHAZ microstructure ....................................................................... 15

5.4. CCT diagram for CGHAZ of BS900D steel ...................................................................... 20

5.5. Charpy impact toughness ................................................................................................... 22

5.6. Verification of microstructure modeling results ................................................................ 23

5.7. Testing applications ........................................................................................................... 24

6. Summary and Conclusions ....................................................................................................... 26

7. Acknowledgements ................................................................................................................... 27

8. References: ................................................................................................................................ 27

Page 3: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

3

Executive Summary

Extra-high strength steels (e.g., BS900D) are increasingly deployed in construction machinery and

heavy manufacturing industries to build strong yet light-weight structures. These advanced steels

have a carefully engineered base metal microstructure, which can be significantly altered by

welding thermal cycles. The end users of extra-high strength steels desire incorporating welding

into their “virtual” manufacturing for product performance optimization to significantly decrease

the time and cost of new product development. However, developing an effective way to model

the weld mechanical properties of extra-high strength steels remains a major technical challenge.

Addressing this technical challenge, the overall goal of this research is to develop the basic

knowledge of weld microstructure and mechanical properties of extra-high strength steels.

BS900D, a hot-rolled extra-high strength steel developed by Basosteel for the construction

machinery industry, is chosen for the study. It is focused on the coarse-grained heat affected zone

(CGHAZ), the most critical region in a welded joint.

Gleebe based physical simulation is used to produce samples with bulk CGHAZ microstructure

under different cooling rates. Microstructure models are established based on the experimental

data of prior austenite grain size, hardness and Charpy impact toughness. The microstructure

models are incorporated into a finite-element based weld heat transfer model. This integrated

model is capable of calculating the weld temperature distribution and mechanical properties for

various arc welding processes (e.g., gas tungsten arc welding, gas metal arc welding and

submerged arc welding) over a wide range of heat inputs. The developed framework of integrated

weld modeling provides an experimental and analytical foundation for understanding the CGHAZ

properties of other extra-high strength steels in the future.

Page 4: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

4

1. Introduction

Complex, transient physical processes take place during welding due to the interaction between

the heat source (e.g., arc) and the workpiece material (e.g., high-strength steel).[1,2] These physical

processes include rapid heating and melting of the base metal, vigorous molten metal flow in the

weld pool driven primarily by the surface tension stress (or Marangoni stress) and Lorentz force,

solidification of the molten pool, and subsequent cooling during which various solid-state phase

transformations take place resulting in the final microstructure. The microstructure changes incur

by welding can have detrimental effects on weld properties such as reduction in ductility and

fracture toughness and softening of heat-affected zone (HAZ).

Although the theories for welding-induced microstructure changes are well established and

published in a variety of open textbooks,[3,4] the development of practical solutions (e.g.,

optimizing welding parameters, pre-heating and post-weld heat treatment or PWHT) are

oftentimes obtained through an experimental trial and error approach. As extra-high strength steels

are increasingly deployed in construction machinery and heavy manufacturing industries to build

strong yet light-weight structures, the trial and error approach can be costly in term of time and

resources required to conduct all the experiments. To address this challenge, many computation

codes (both commercial and open source) have been developed to calculate the evolution of

temperature and microstructure during welding. In particular, with the widespread of computed

aid engineering (CAE) tools, the end users of extra-high strength steels desire incorporating

welding into their “virtual” manufacturing for product performance optimization.

Extra-high strength steels have a carefully engineered base metal microstructure, which is

significantly altered by welding thermal cycles. Computational simulation of microstructure

evolution during welding and resulting mechanical properties is still evolving. In particular, as the

final weld microstructure can strongly depend on the chemistry and the initial microstructure, the

existing microstructure models are typically limited to some particular chemical compositions of

the steels for which those models are calibrated. Developing an effective way to model the weld

mechanical properties of extra-high strength steels remains a major technical challenge.

2. Research Objectives

Addressing this technical challenge, the overall goal of this research is to develop the basic

knowledge of weld microstructure and mechanical properties of extra-high strength steels.

BS900D, a hot-rolled extra-high strength steel developed by Baoshan Iron and Steel Research

Institute (hereafter referred as Baosteel for brevity) for the construction machinery industry, is

chosen for the study. It is focused on the coarse-grained heat affected zone (CGHAZ), the most

critical region in a welded joint.

The specific objectives and tasks are:

Gleeble physical simulation

o Dilatometry for continuous cooling transformation (CCT)

o Microstructure characterization and hardness testing

Page 5: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

5

o Charpy impact testing

Integrated weld modeling:

o Weld process model

o Microstructure model of hardness and Charpy toughness (-40 °C and room

temperature)

Testing applications

An overview of the research tasks is illustrated in Figure 1. In particular, the Gleeble physical

simulation was used to generate the experimental data under controlled heating and cooling

conditions. Such experimental data was used to develop empirical equations for describing weld

microstructure including prior austenite grain size, hardness and Charpy impact toughness. It is

noted that the individual phase fractions (e.g., those of bainite and martensite) were not calculated.

As discussed in details later, this is because that the tested CGHAZ of BS900D steel comprised a

mixture of bainite and martensite that had similar morphology and hardness and were difficult to

quantitatively differentiate from each other. The empirical equations were integrated with a weld

heat transfer model based on Abaqus finite element analysis (FEA) code to predict the joint

mechanical properties. The integrated weld model was tested with experimental data of

autogenous gas tungsten arc welding (GTAW).

Figure 1: Overview of research tasks.

Although the model was tested for autogenous GTAW, the integrated weld model is capable of

considering other commonly used welding processes such as gas metal arc welding (GMAW) and

submerged arc welding (SAW) and a wide range of welding heat inputs. Moreover, the framework

developed can be expanded to other extra-high strength steels in the future.

Page 6: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

6

3. Experimental Approaches

As discussed earlier, the base metal studied is BS900D, an extra-high strength steel. The

composition and the mechanical properties of base metal are summarized in Table 1.

Table 1: Chemical composition and mechanical properties of BS900D base metal

Composition (wt%)

C Si Mn Cr Mo Ti Nb B V Alt P S

BS900D 0.16 0.22 1.2 0.28 0.29 0.018 0.016 0.0012 0.04 0.04 0.0114 0.0029

Hardness: 325 Vickers

Tensile strength: 965 MPa

Impact toughness: 65 J (longitudinal direction) 43 J (transverse direction)

3.1. Gleeble-based physical simulation of CGHAZ microstructure

The samples used for Gleeble physical simulation were machined from a 8.5-mm-thick BS900D

plate provided by Baosteel. As shown in Figure 2, the rectangular bar shaped samples (dimensions

11 × 6 × 100 mm) were cut perpendicular to the rolling direction of the plate.

Figure 2: Samples used for Gleeble physical simulation of CGHAZ microstructure.

A Gleeble® 3800 thermomechanical tester at OSU was used for the physical simulation of

CGHAZ microstructure. The experimental setup for Gleebe physical simulation is shown in

Figure 3. The sample was placed between a pair of water cooled copper grips which, in turn, were

attached to a pair of low-force jaws. A controlled resistive heating was used to heat the sample in

which the low-force jaws permitted the free expansion and contraction of the sample arisen due to

thermal expansion. The thermal cycle to induce a CGHAZ microstructure consisted of a heating

rate of 85°C/sec to a peak temperature of 1350°C, holding at peak for 1 and final “free” cooling

down to room temperature. The cooling rate was varied by changing the free span between the

pair of copper grips from 10 to 50 mm; the shorter the free span, the faster the cooling rate. A

dilatometer was placed on the rectangular sample to record the volume change on heating cooling.

It is noted that the free cooling was used to more accurately determine the phase transformation

temperatures from the dilatometry curve based on the deviation in slope (i.e., coefficient of thermal

expansion).

Page 7: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

7

Figure 3: Experimental setup for Gleeble physical simulation.

Three thermocouples (TCs) were mounted on the sample surface along the axial direction: one at

the center and the other two at each side of the center TC. The TC data showed that the gauge

section (i.e., length of sample exposed to ± 15 °C of peak temperature) was about 6 mm long at

the center of sample.

3.2. Samples for Charpy impact testing

The second batch of Gleeble physical simulation was used to produce samples with bulk CGHAZ

microstructure for Charpy impact testing. For improve consistency in the results, “controlled”

force cooling (not free cooling) was used to maintain the specified constant cooling rates for a free

span fixed at 10 mm. The peak temperature was also 1350 °C. In addition to 1 s hold at peak, a

longer hold time of 10 s was used to approximate the reheated CGHAZ with large prior austenite

grains.

3.3. Charpy impact testing

After the Gleeble physical simulation, the center portion of a sample was machined into a subsize

specimen for Charpy impact testing, as shown in Figure 4. The geometry of subsize Charpy

specimen (dimension 10555 mm) was prepared in accordance to the ISO 148 and ASTM A370

standards. It is noted that the dimensions of the subsize specimen are the same as those used in

the prior Charpy testing of base metal performed by Baosteel. Charpy impact testing was

performed in accordance to ASTM Standard E23 - 12c, “Standard Test Methods for Notched Bar

Impact Testing of Metallic Materials” at two testing temperatures: -40 ℃ and 20 ℃ (room

temperature).

Page 8: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

8

Figure 4: Schematics showing the subsize Charpy specimen machined from Gleeble

sample.

3.4. Microstructure characterization

The Gleeble heat treated samples were prepared for microstructure characterization by following

the standard metallographic procedure. The sample was etched in a 5% Nital solution for

observing the bainite and martensite microstructure using optical microscope and scanning

electron microscope (SEM).

The prior austenite grain size is a critical microstructure parameter as it can influence the kinetics

of austenite decomposition into different ferrite micro-constituents including coarse martensite-

austenite (M-A) constituent, which has been linked to reduced impact toughness. Multiple etching

methods were tested to reveal the prior austenite grain boundary. The best method for etching the

prior austenite grain size in CGHAZ of BS900D was found to be:

Step 1: Vilella’s Reagent (1% picric acid + 5% HCL) for ~90 sec

Step 2: A light polish (1 micron) with normal pressure for ~30 sec

Step 3: 5% Nital for ~10 s

ImageJ, an open source image processing software, was used for grain size analysis. For each

testing condition, two to three images were analyzed at 500x magnification. The freehand tool

was used to outline and measure the grain boundary areas of 13-20 grains in the CGHAZ region

to determine the average grain size. Some measurements were further confirmed on SEM images.

Finally, the macro-hardness was measured using a Leco indentation machine with 1 kg load. To

evaluate the microstructure-dependent hardness, the micro-hardness measurement was done using

another LECO Microhardness Tester LM100AT with 300 gram load.

4. Approaches for Integrated Weld Modeling

4.1. Equation for grain growth in CGHAZ

The classic grain growth theory indicates that the driving force for grain growth is the decrease of

interface energy of grains and the kinetics is controlled by diffusion. Under isothermal condition,

the rate of grain growth can be described as:

Page 9: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

9

tRT

QkDD ann

exp/1

0

/1 (1)

where D is the mean grain size (μm), D0 is the initial grain size (μm), k is a kinetic constant, Qa is

the activation energy, n is the growth exponent, R is the universal gas constant, T is the temperature

(°C), and t is the time (s).

During welding, the grain growth occurs under non-isothermal condition of the rapid weld heating

and cooling. For this case, Eqn. (1) can be rewritten in the following integration form:

dttRT

QkDD

tann

0

/1

0

/1

)(exp (2)

where T(t) is the temperature profile as a function of time. Eqn. (2) represents the summation of

grain growth over many small isothermal increments to obtain the final grain size.

The austenite grain growth is assumed to occur at a temperature above the austenitizing

temperature Ac3. This start temperature of grain growth is chosen to be 900 °C. Moreover, the

peak temperature for grain growth is capped at 1350 °C, above which the austenite starts to

transform into -ferrite phase.

In Eqn. (2), there are three material parameters: n, Q and k. These parameters were determined

using regression analysis of the experimental data of grain size obtained from Gleeble physical

simulation. The final equation for grain growth in CGHAZ of BS900D is given as:

dttTtT

Dr

P

P

c

t

t

t

t

)(

25920exp

)(

25920exp102.7625.11 1238.238.2

(3)

where n = 0.42, k = 2.76×1012, and R

Q=25920. The initial austenite grain size D0 is taken the same

as that of the base metal (11.25 µm). tc is the time reaching the grain growth temperature

(900 °C) during heating, tp is the time reaching the peak temperature, and tr is the time reaching

the phase transformation temperature (also taken as 900 °C) during cooling. The first integration

term on the right hand side of Eqn. (3) is the grain growth during heating and the second term is

the grain growth during cooling.

4.2. Equation for CGHAZ hardness

Hardness equations developed by R. Blondeau et al.[5] have been widely used for weld

microstructure modeling. For instance, it was used by Ion et al. for developing diagrams of

microstructure and hardness for HAZs in welds.[6] Those equations, taking into account the base

metal composition and weld cooling rate, have the following general form:

rHVHV VbaHV 10log (4)

where HV is the hardness of a micro- constituent (i.e., ferrite, bainite or martensite), aHV and bHV

are simple, linear functions of steel composition, and Vr is the cooling rate at 700 °C (in °C/hour).

Page 10: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

10

As discussed in details later, the tested CGHAZ of BS900D steel comprised a mixture of bainite

and martensite with similar morphology and hardness. It was difficult to quantitatively

differentiate one micro-constituent from another. Since the microstructure is predominantly

bainite, the model for CGHAZ of BS900D followed the general form of Blondeau’s hardness

equation for bainite. Two constants, aHV and bHV, were determined by regression analysis of

experimental hardness data measured on the Gleeble samples. The final equation of CGHAZ

hardness for BS900D is given as:

rV

HV

10log33Mo–20Cr–10Ni–22Mn–55Si–53C156.9

190Mo144Cr65Ni153Mn330Si185C 547.8–

(5)

where the concentrations of alloying elements such as C, Si and Cr are given in weight percent

(wt%).

4.3. Equation for Charpy impact toughness of CGHAZ

Although this is no lack of experimental data of Charpy impact toughness for CGHAZ in various

steel welds,[7] a generally-accepted equation to predict the CGHAZ toughness is yet to be

established. Two parameters have been recognized for their effects on the Charpy impact

toughness: cooling rate and prior austenite grain size. The effect of cooling rate is expected as it

influences the microstructure formed in the CGHAZ. For the prior austenite grain size, it has been

shown in the literature that the coarsened austenite grains can form a high fraction of M-A

constituent. The formation of hard M-A constituent has a detrimental effect on the impact

toughness, especially at low temperatures (below -20 °C).

In this study, the equation for Charpy toughness of CGHAZ has the following simple form:

DctbaJ JJJT 5/8 (6)

where JT is the Charpy toughness at temperature T (either -40 °C or 20 °C), t8/5 is the cooling time

from 800 to 500 °C, and D is the prior austenite grain size. The coefficients, aJ, bJ and cJ, are

determined based on regression analysis of experimental data of Charpy toughness. The final

equations for Charpy toughness are given by the following equations:

γ/C DtJ 0.051 – 0.305 – 43.75 5820 at +20 °C (7)

γ/C DtJ 0.025 – 0.113 – 16.80 5840 at -40 °C (8)

4.4. Welding heat transfer model

The governing equation for heat transfer is given as:

2

2

2

2

2

2

z

T

y

T

x

Tk

t

TCP (9)

where is the density (kg/m3), CP is the specific heat (J/kg-K), T is the temperature (K), t is the

time, k is the thermal conductivity (J/m-s-K), and x, y and z are the coordinates (m).

In the weld pool, the heat transfer is significantly enhanced by the molten pool flow.

Computational modeling of molten pool flow is complex and is typically solved using the

Page 11: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

11

computational fluid dynamics (CFD) method. For FEA-based welding simulation (such as Abaqus

and Sysweld), a so-called double ellipsoid equation, developed by Goldak et al.,[8] has been widely

used. This equation describes a volumetric heat flux distribution from the welding arc as:

2

2

2

2

2

2 333

36),,(

c

z

b

y

a

x

eeeabc

Qzyxq

(10)

where q(x,y,z) is the heat flux (J/s-m3) at a point with a distance x, y and z to the center of the

welding arc, Q is the heat input where Q = * I * V (efficiency * current * voltage), and a, b, c

are the three heat source parameters.

Using the double ellipsoid equation, the weld temperature field has been satisfactorily calculated

using the FEA approach for a variety of welding processes such as GMAW, SAW and GTAW. It

is noted that as the double ellipsoid equation does not take into account the weld pool physics, the

three heat source parameters need to be calibrated with a weld macrograph. In general, a is taken

as front/rear length of weld pool, b as the half width, and c as the depth of the weld pool. If

available, temperature profiles measured by TCs at locations close to the weld metal are desirable

to validate the weld heat transfer model using the double ellipsoid equation.

4.5. Integrated weld modeling

The flowchart of the integrated weld modeling is shown in Figure 5. The main program is the

Abaqus-based weld heat transfer model. First, a mesh comprising nodes and elements is created

in Abaqus CAE (a FEA pre- and post-processor) for the given weld geometry. Second, the input

file is generated including the mesh and thermal-physical properties of the steel. Third, the input

file is solved using Abaqus Standard (a FEA solver) where the welding parameters are taken into

account through Abaqus user subroutine coded in Fortran. This heat transfer model calculates the

temperature profiles in the weld joint.

Figure 5: Flowchart of integrated weld modeling.

Page 12: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

12

The calculated temperature profiles are inputted into the weld microstructure coded in Python,

which can access the temperature data directly from the Abaqus output database (odb file). It

calculates the prior austenite grain size by integrating Eqn. (3) over the temperature profile. The

hardness is calculated using Eqn. (5) by inputting the cooling rate at 700 C determined from the

temperature profile. Finally, the Charpy toughness at -40 C and 20 C is calculated using Eqns. (7)

and (8) by inputting the calculated prior austenite grain size and the cooling time t8/5 determined

from the temperature profile. The calculated results can be visualized in Abaqus CAE. Additional

details of the integrated weld model are provided in the accompanying user manual.

5. Results and Discussion

5.1. Welding thermal simulation in Gleeble

As discussed previously, two batches of Gleeble samples were produced. The first batch was used

to simulate the CGHAZ microstructure and observe the phase transformation temperatures. A free

cooling condition was utilized for the first batch. On the other hand, the second batch was used to

produce samples that were subsequently machined for Charpy impact testing. For the second

batch, a controlled cooling condition was utilized to maintain consistent cooling rate.

The actual temperature profiles for Gleeble physical simulation of CGHAZ microstructure are

shown in Figure 6. As discussed previously, the hold time at the peak temperature 1350 °C was

1 s. The free span between the copper grips has a significant effect on the cooling rate. The

cooling time from 800 to 500 °C t8/5 reduces from 44.4 s for a free span of 50 mm only 3.8 s for

a free span of 10 mm.

Figure 6: Actual temperature profiles in Gleeble physical simulation of CGHAZ

microstructure.

0

200

400

600

800

1000

1200

1400

0 50 100 150 200 250

Tem

per

atu

re (°C

)

Time (sec)

FS = 10 mm

t8/5 = 3.8 sec

FS = 20 mm

t8/5 = 7.7 sec

FS = 30 mm

t8/5 = 14.8 sec

FS = 40 mm

t8/5 = 22.2 sec

FS = 50 mm

t8/5 = 44.4 sec

Page 13: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

13

The cooling portion of the temperature profile exhibits a small plateaus, which is likely caused by

the release of latent heat by austenite to ferrite transformation. Using the initial point of this

plateaus as the approximate start temperature of phase transformation, a higher undercooling is

needed for the austenite transformation the cooling rate is increased, a phenomenon commonly

observed in low alloy steels.

Figure 7 shows the actual temperature profiles for Gleeble simulation of Charpy specimens. Three

cooling times t8/5 were evaluated: 7.7, 14.8 and 22.2 s, where a hold time of 1 s was used. For the

cooling time t8/5 = 7.7 s, an extended hold time of 10 s was used to approximate the coarsened

austenite grains in the reheated CGHAZ. As shown in this figure, Gleeble was able to maintain a

steady cooling rate until 400 °C, below which the cooling rate was slowed.

Figure 7: Actual temperature profiles for Gleeble simulation of Charpy specimens.

5.2. Base metal microstructure

As shown in Figure 8, the BS900D base metal microstructure is mainly composed of granular

bainite (BG) and lath bainite/martensite (BL). The microstructure appears to be fairly equiaxed

with an average diameter of approximately 11.25 μm. Higher magnification SEM images of the

base metal microstructure are shown in Figure 9. In addition to BG and BL, a small amount of

proeutectoid ferrite (designed as F) is observed along original austenite boundary.

Page 14: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

14

Figure 8: Base metal microstructure observed in optical microscope.

As shown in Figure 9, uniformly-dispersed fine precipitates inside granular bainite are observed.

Given the composition of BS900D shown in Table 1, these precipitates are likely carbides formed

with the micro-alloying elements such as titanium, vanadium and niobium. The fine precipitates

can act as nucleation sites during phase transformation to promote grain refining. Such

microstructure comprising fine grains and dispersed precipitates is likely essential to the balanced

mechanical properties of extra high strength (tensile strength 960 MPa) and decent toughness at

low temperature (e.g., Charpy toughness 43 J transverse direction at -40 °C). The average hardness

is 325 HV taken from six Vickers hardness indents at a 1 kg load.

Figure 9: SEM images of base metal microstructure.

Page 15: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

15

During welding, the carefully-engineered baes metal microstructure is significantly altered by

welding thermal cycles, as discussed in the following section.

5.3. Gleeble-simulated CGHAZ microstructure

The CGHAZ microstructure simulated in Gleeble is shown in Figure 10. It comprises a mixture

of martensite and bainite. In particular, it transitions from a higher fraction of fine lath martensite

at the fastest cooling to a higher fraction of granular bainite at the slowest cooling.

Figure 10: Simulated CGHAZ microstructure observed in optical microscope.

Higher magnification SEM images of the fine lath martensite, interlaced martensite and granular

bainite are shown in Figure 11(a), (b), and (c), respectively.

The martensite and bainite micro-constituents can be identified by carefully examination of the

high-resolution SEM images. However, the morphology of the two micro-constituents is similar,

especially when observed in low magnification optical images. Such similar morphology makes

it difficult to quantitatively determine the fraction of each phase.

Page 16: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

16

Figure 11: SEM images of CGHAZ microstructure simulated in Gleeble.

As discussed earlier, a two-step etching method, which comprised the first etchant of Vilella’s

Reagent followed by the second etchant of 5% Nital, was found to be effective in revealing the

prior austenite grain size in CGHAZ of BS900D. Example optical and SEM images are shown in

Figure 12, where the prior austenite grain boundary can be readily observed.

Figure 12: Example images of prior austenite grain boundary revealed by the two-step

etching method.

Figure 13 shows the optical images of austenite grain growth under different cooling conditions

and hold times. As expected, the longer the hold time and the longer the cooling time (slower

cooling rate), the larger the prior austenite grain size. Such grain growth in CGHAZ occurs in part

due to the dissolution of Nb, V and Ti-rich precipitates, which were uniformly-dispersed in the

original base metal. Although still significant, the grain growth in BS900D is much less than that

in a HSLA steel (about 200 µm) under comparable thermal cycle.[9]

The prior austenite grain size is further summarized in Table 2. It is shown that the free and

controlled cooling conditions at a given t8/5 time result in almost identical grain size. This is

because the majority of austenite grain growth takes place at elevated temperature (say above

Page 17: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

17

900 °C). As a result, the difference in temperature profile below 900 °C between the free and

controlled cooling conditions does not have a significant effect on the prior austenite grain size.

Figure 13: Austenite grain growth under different cooling conditions and hold times.

Page 18: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

18

Table 2: Prior austenite grain size as a function of cooling time.

Cooling Hold

Time (s) t8/5 (s) CR (°C/s)

Avg. size

(μm)

Max. size

(μm)

Min. size

(μm)

Stdev

(μm)

Free cooling 1 3.8 78.7 57 71 46 7

Free cooling 10 4.4 67.7 116 161 76 27

Free cooling 1 7.7 39.0 66 90 51 10

Controlled 1 7.7 39.0 64 102 55 10

Controlled 10 7.7 39.0 116 175 76 25

Free cooling 1 14.7 20.4 71 95 55 10

Controlled 1 14.7 20.4 76 89 61 8

Free 1 22.2 13.5 85 118 48 15

Controlled 1 22.2 13.5 88 118 66 16

Free 1 44.4 6.8 97 154 56 23

The micro-hardness measurements of the Gleeble simulated samples under the fastest and slowest

cooling rate tested are shown in Figure 14 and Figure 15, respectively. Micro-hardness for other

cooling conditions shown in Table 2 were also measured. The individual plots are not shown here

for brevity but the average micro-hardness as a function of cooling rate will be summarized later.

As expected, the micro-hardness is the highest under the fastest cooling rate. The hardness

decreases as the cooling rate is reduced. This is consistent with the microstructure of the CGHAZ

discussed earlier, transitioning from a higher fraction of fine lath martensite at the fastest cooling

to a higher fraction of granular bainite at the slowest cooling

Although there is significant variation in hardness as a function of cooling rate, it is difficult to

observe much microstructure-specific hardness variation under a given cooling rate. This is one

reason that the hardness model for CGHAZ of BS900D steel was developed for the mixture of

bainite and martensite as shown in Eqn. (5).

Page 19: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

19

Figure 14: Micro-hardness of a Gleeble simulated sample under the fastest cooling rate

tested (t8/5=3.8 s). The average hardness is 431 Vickers.

Page 20: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

20

Figure 15: Micro-hardness of a Gleeble simulated sample under the slowest cooling tested

(t8/5=44.4 s). The average hardness is 324 Vickers.

5.4. CCT diagram for CGHAZ of BS900D steel

Figure 16 plots the dilatometry curves of Gleeble simulated samples cooled at the fastest (t8/5 =

3.8 s) and slowest (t8/5=44.4 s) cooling rates. Both samples were heated to the peak temperature

of 1350 °C, held for 1 s and then free cooled. The start and finish temperatures of the austenite to

ferrite transformation are readily determined based on slope analysis.

Figure 16: Dilatometry curves of Gleeble simulated samples cooled at the fastest and

slowest cooling rates. The phase transformation start and finish temperature are marked.

Page 21: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

21

The start and finish temperatures of austenite to ferrite transformation during cooling are plotted

in Figure 17 as a function of cooling rate. As expected, the faster the cooling rate, the lower the

transformation temperatures. When the cooling time t8/5 is shorter than 7.7 s (i.e., cooling rate =

39 °C/s), the change in transformation temperature with respect to cooling time is relatively small,

which is consistent with an austenite to martensitic transformation.

Figure 17: Start and finish temperatures of austenite to ferrite transformation during

cooling as a function of cooling rate

Figure 18 plots the CCT diagram for CGHAZ of BS900D steel. The Gleeble measured start and

finish temperatures are plotted as red diamonds and yellow triangles, respectively, superimposed

on the data calculated by JMatPro, a commercial microstructure code. The JMatPro results are

used to illustrate the phase transformation under very slow cooling rates (less than 7 °C/s).

Although such slow cooling rates can be readily tested in Gleeble, they are expected to be much

slower than the practical cooling rates encountered in welding.

The JMatPro results indicate A1 temperature equal to 718 °C and A3 equal to 837 °C, which are

consistent to the Gleeble measurement (Ac1 = 734 ℃ and Ac3 = 845 ℃). For transformation start

temperature, the JMatPro results of the bainite start line are consistent with the Gleeble

measurement for cooling times from 14.8 to 44.4 s. The transformation start temperatures at

shorter cooling times (t8/5 = 7.7 and 3.8 s) measured by Gleeble are slightly higher than the

martensite start line predicted by JMatPro. On the other hand, the consistence for the

transformation finish temperature is low between the JMatPro results and the Gleeble

measurement. Finally, the JMatPro results show a fairly gradual decrease in hardness from 409 to

389 Vickers as the cooling time increases from 3.8 to 22.2 s. The experimental data has a much

higher decrease in hardness from 424 to 342 Vickers.

Such inconsistence in transformation finish temperatures and hardness indicates the importance of

validating the microstructure prediction using experimental data.

Page 22: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

22

Figure 18: CCT diagram for CGHAZ of BS900D steel. The Gleeble measured start and

finish temperatures are plotted as red diamonds and yellow triangles, respectively, on the

data calculated by JMatPro.

5.5. Charpy impact toughness

The Charpy impact toughness of Gleeble simulated CGHAZ is plotted in Figure 19. For each

condition, there were four repeats tested. The average of the four tests is plotted as the solids

whereas the error bar indicates the range (lower and upper limits) of the four tests.

Figure 19(a) shows the Charpy impact toughness as a function of t8/5, where the prior austenite

grain size for each condition is labeled. Figure 19(b) plots the Charpy impact toughness as a

function of prior austenite grain size for the same cooling time t8/5 = 7.7 s. As shown in this figure,

for the limited conditions tested, the impact toughness at either temperature is not sensitive to the

cooling time and the prior austenite grain size.

One possible explanation for the relatively stable Charpy impact toughness is the counteracting

effect of grain size and micro-constituent. When the cooling rate is fast, the microstructure is

largely fine lath martensite, as shown in Figure 11. At slow cooling rate, bainite forms first.

Although bainite is expected to have better toughness than martensite in principle, the coarsened

grains and the formation of M-A constituent may have a detriment effect on toughness.

Page 23: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

23

Figure 19: Charpy impact toughness of Gleeble simulated CGHAZ at -40 and 20 °C. For

clarify of display, the data points at the shortest cooling time (t8/5=7.7 s) are offset.

5.6. Verification of microstructure modeling results

Physics-based modeling of weld microstructure remains a major challenge due to the complexity

in non-equilibrium phase transformation.[7] As discussed previously, the microstructure models

were established by regression analysis of experimental data for Gleebe simulated CGHAZ of

BS900D steel.

To verify the accuracy of the regression analysis, the modeling results are compared to the

experimental data (which was used in the regression analysis). As shown in Figure 20, the models

of prior austenite grain size, hardness and Charpy impact toughness can reasonably fit the

respective experimental data.

Figure 20: Verification of microstructure modeling results (a) prior austenite grain size, (b)

hardness, and (c) impact toughness.

Page 24: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

24

The validity of the microstructure models is evaluated in the testing application when the

integrated weld model is applied to an actual weld of GTAW, as discussed in the next section.

5.7. Testing applications

To validate the integrated weld model, an actual GTA weld of BS900D plate was evaluated. The

welding parameters are summarized in Figure 21. The CGHAZ microstructure is similar to that

obtained in Gleelbe simulation, as expected.

Figure 21: GTA welding of BS900D plate for validating the integrated weld model.

Figure 22 compares the calculated and measured weld metal profile. The comparison between the

calculated and measured prior austenite grain size is shown in Figure 23(a). The maximum grain

size calculated by the model is 98 µm, which is fairly consistent to the measured grain size of

107 µm. Figure 23(b) compares the calculated and measured hardness for CGHAZ. The peak

hardness of 350 Vickers calculated is somewhat higher than the experimental data of 300 Vickers.

Such over-prediction of hardness likely caused by lack of calibrating the heat source parameters

in Eqn. (10).

Page 25: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

25

Figure 22: Comparison between the calculated (left) and measured (right) weld metal

profile.

Figure 23: Comparison between calculated and measured prior austenite grain size and

hardness for CGHAZ of BS900D steel.

Although the above testing application is for GTAW, the integrated weld model developed here is

capable of modeling other welding processes such as GMAW and SAW over a wide range of

welding heat inputs. The only change in the model that is needed is the adjustment of heat source

parameters in the double-ellipsoid heat flux for the respective welding process.

Page 26: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

26

To demonstrate this, a double-sided joint welded by GMAW was tested. The experimental data

was taken from the welding work of BS960QC by Qu et al.[10] Figure 24 shows the predicted

results of prior austenite grain size, hardness and Charpy impact toughness at -40 °C.

Figure 24: Testing application of integrated weld model to GMAW of BS960QC.

6. Summary and Conclusions

The basic knowledge of weld microstructure and mechanical properties of extra-high strength steel

BS900D is established based on a combination of Gleeble physical simulation and computational

modeling. The following conclusions can be drawn:

The CGHAZ microstructure of BS900D transitions from a higher fraction of fine lath

martensite at fast cooling to a higher fraction of granular bainite at slow cooling.

The longer the hold time and the longer the cooling time (slower cooling rate), the larger

the prior austenite grain size, as expected. Although still significant, the grain growth in

BS900D is much less than that in HSLA steel (about 200 µm) under comparable thermal

cycle.

For the limited conditions tested, the impact toughness at -40 and 20 °C is not sensitive to

the cooling time and the prior austenite grain size. One possible explanation for the

relatively stable Charpy impact toughness is the counteracting effect of grain size and

micro-constituent: fine martensite at fast cooling versus coarse bainite at slow cooling.

Equation of prior austenite grain growth is established based on the classic grain growth

theory. A time integration scheme is used to calculate the grain growth under non-

isothermal heating and cooling condition in welding.

Page 27: Basic Research of Weld Process and Microstructure Modeling for … · 2019-12-16 · 1 Basic Research of Weld Process and Microstructure Modeling for a Hot-Rolled High Strength Steel

27

The hardness equation is developed based on the widely used formula by Blondeau et al.

The bainite and martensite in the CGHAZ microstructure of BS900D exhibit similar

morphology and hardness, making it difficult to quantitatively determine the phase

fractions. As a result, the hardness as a function of cooling time is established by

considering a mixture of bainite and martensite (as opposite to the hardness of individual

phases).

The Charpy impact toughness model is established based on a simple bi-linear relationship

of prior austenite grain size and cooling time t8/5.

An integrated weld modeling framework is developed for CGHAZ of BS900D steel. It is

capable of modeling several arc welding processes such as GTAW, GMAW and SAW over

a wide range of welding heat inputs. The only change in the model that is needed is the

adjustment of heat source parameters in the double-ellipsoid heat flux for the respective

welding process.

7. Acknowledgements

This fundamental research was supported financially by Baosteel Research Institute in Shanghai,

China. The authors are grateful to Mr. Sun Zhongqu, Mr. Qu Xueyuan, and Dr. Zhaoxia Qu and

her team for their valuable guidance on the research.

8. References:

1 K. Easterling: Introduction to the Physical Metallurgy of Welding, Butterworth-Heinemann,

Oxford, 1992. 2 S. A. David and T. DebRoy: Science, Vol. 257, p. 497 (1992). 3 S. Kou: Welding Metallurgy, John Wiley & Sons, Hoboken, New Jersey, 2003. 4 J. C. Lippold, S. D. Kiser and J. N. DuPont, Welding Metallurgy and Weldability of Nickel-Base

Alloys, 2009. 5 R. Blondeau et al., Heat Treatment 76, Metals Sot., London (1976). 6 Ion et al., A second report on diagrams of microstructure and hardness for heat-affected zones

in welds, Acta Metallurgica, Vol. 32, pp. 1949-1962 (1984). 7 Ø. Grong: Metallurgical Modeling of Welding, 2nd edition, The Institute of Materials, London,

1997. 8 Goldak et al., A new finite element model for welding heat sources, Metallurgical

Transactions B, Vol. 15, pp. 299-305 (1984). 9 R. Cao et al., Micromechanism of Decrease of Impact Toughness in Coarse-Grain Heat-Affected

Zone of HSLA Steel with Increasing Welding Heat Input, Metallurgical and Materials Transactions

A, Vol. 46a, p. 2999 (2015). 10 Z. Qu, H. Wang and L. Xu, Research on welding technology of the hot-rolled extra-high

strength steel developed by Baosteel, Baosteel Technical Research, Issue 3, pp. 3-9, 2013