(1988). J. J. Jonas and I. Weiss: Mel. Sci., ... H. Stuart,. H. Zhang and C. Li,. China. Science &Technology Press, Beijing, (1995), 179. 31). L. P. Karjalainen.
ISIJ International,
Vol.
35
(1
995). No, 12, pp.
1523-1 531
Softening and Flow Stress Behaviour of Nb Microalloyed Steels during Hot Rolling Simulation
L.
Pentti
Terrence M. MACCAGNO1) and John J. JONAS1) KARJALAINEN.
Department of Mechanical Engineering. University of Oulu, P.O. Box440,
1) Department of
Metallurgical
(Received on April
Engineering, McGil[ University, lO. 1995.•
accepted in
final
Fl N-90571 Oulu, Finland. E-mail: penttik@me,oulu.fi 3450 University Street, Montreal. CanadaH3A2A7.
form on
May26.
1995)
The roles of softening and precipitation were investigated by meansof hot torsion experiments under conditions simulating either plate or sheet rolling. Six microalloyed steels containing Nbwere studied. During the first few finishing passes in the sheet rolling simulations, the meanflow stress (MFS) increased as the interpass time wasdecreased. Dueto strain accumulation, the rate of static recrystallization (SRX) increased after each pass. By taking both strain accumulation and grain Tefinement into account, it is significantly shown that SRXplays a marked role under sheet rolling conditions, even at temperatures below the no-recrystallization temperature for plate rolling conditions. The accumulated or retained strain reaches the critical value required to initiate dynamic recrystallization only at the lowest entry and ro]Iing temperatures and shortest interpass times. The kinetics of the strain-induced precipitation of NbCNunder continuous cooling conditions, taking partial SRXinto account, indicate that precipitation begins after to 5 passes when3s interpass times are employed, thus reducing further softening. But when I s interpass times are used, most of the passes take place before copious precipitation, so that static and post-dynamic (i.e. metadynamic) softening maycontinue to take place. As a result, the MFSIevel decreases as the interpass time is shortened during the final passes. The extent of grain refinement wassimilar in both the sheet rolling and plate rolling simulations. The ferrite grain size is shownto depend on MFSof the final pass, and is independent of the chemical composition of the microalloyed steel.
2
hot KEYWORDS: recrystallization;
1.
ro]ling
simulation,
torsion testing;
dynamic recrystallization;
Nb microalloyed
al.3~5)
Introduction
Nb-V
Various rolling schedules can be utilized to produce high strength low alloy steel strip with fine ferrite grain sizes. However, accurate knowledge of the rolling loads is important for good gauge and flatness control of the product, as well as for the design of optimum pass schedules. This in turn requires an understanding of the kinetics of carbonitride precipitation, as well as of the static
and dynamic restoration
controlled rolling leads to the finest ferrite grain sizes and lowest rolling 10ads,i'3) Differences have also been reported between various microalloyed steels concerning their dynamic
(SRX)
static
found that DRXdoes not take place at all in under tube finishing conditions. Recently,
metadynamic recrystallization
(MRX) fol-
The aim of the present work was to account for the evolution of the flow stress in Nb microalloyed steels under simulated multipass hot rolling conditions. Six different grades were tested using torsion simulations of typical plate and sheet rolling schedules. The evolution of the flow stress was recorded and the final ferrite grain sizes were determined. The kinetics of softening and precipitation under multiple-pass deformation and continuous cooling conditions were analysed in more detail than in previous investigations to clarify the role of the softening mechanisms,as well as the extent to which interferes with softening in Nb steels. precipitation 2.
Experimental
Lowcarbon (0.05
recrystallization al.i•2)
grain size; static
NbCN
DRX,
dynamic/metadynamic recrystallization
(DRX) and
stress;
precipitation.
lowing has been discussed in connection with sheet processing.9'10) and tube
processes.
behaviours. For example, Samuel et
f]ow
steels
the role of
Torsion testing is often used to simulate multipass hot rolling because of the large strains that can be attained. The flow stresses and microstructures developed under different simulated rolling conditions have been comprehensively investigated by meansof this technique in a series of recent papers.1~10) This work shows that
recrystallization
steels;
metadynamicrecrystallization,
and
a/.7) observed that DRXcan occur in Nb under sheet rolling conditions (characterized by relatively short interpass times), whereas Pussegodaet
Bowdenet
steels
1523
IO
wto/o C) steels containing Nb to O. additions were tested. Twoof these were commercial thermomechanically processed high strength steels, and the others were experimental heats. All were obtained from Rautaruukki Oy, Finland. The
plus other alloying
C 1995 ISIJ
ISIJ International,
Table
Nb-T i-V
Si 0.2-0.3
0.08 0.090 0.050 0.07
l .46 l .5 l l .46 l .57
0.04 0.033 0.041
0.001 0.0
O.04 l
0.0
0.046 O.076
l.51
0.0
l .20
0.046
l
Nb-Ti-Ni low Nb-Ti high Ti-Nb *
Chemical compositions of the experimental
Nb
l
Nb
NbTi
35 (1995), No. 12
Mn
C
Grade
l.
Vol.
15
0.015
1l
18
0.020 O. 142
-
0.005 0.006 0.009 0.005 0.008 0.003
o.70 0.04 0.03
l
Nb-Ti
Nb Nb-Ti-V
160
v,
INb-Ti
LAJ
Qe
H
tl)
:;
O J
120 Tnr
LL
80
z UJ
=
8 = 0.3
s~1
8= 2 = 30 s
40
ti
o O6S
O75
0.85
0.95
1.05
1.1 5
(FC1)
1OOO/T MFSvs,
l
Fig.
the l/T curves for the plate rolling
simula-
tions,
Table
2.
Grade
Measuredand calculated no-recrystallization and transformation start (A*3) temperatures TNr meas.
Nb
Nb-Ti-V
Nt~TiNi Iow Nb-Ti high Ti-Nb high Ti-Nb
Tnr calc.*
Ar3 meas.
980 956 990 945 84 l
945 925 935 915 860 960 985
Nb-Ti
755 760 770 705 740 780
l 135
(Tnr) ('C).
Ar3 calc.**
766 759 774 718 772 780
(=0.2s~1) *
T~,=887 +464c+ 6445Nb-644~Nb)+ (732v-230 ~v
** A,3
(
890Ti + 363Al 357si 9ro 3roc 80Mn-20cu-
=
two of the
-
examplesof the evolution of the mean
flow stress (MFS) during rolling are plotted against the inverse of the absolute pass temperature, I/T. Here, is the area under each stress-strain curve divided by the strain. This type of plot can be used to determine pass
MFS
1524
-
15cr-
55Ni
)+
- 80MO
temperatures of hot rolling: the notemperature, T~*, and the start of the transformation temperature A,3 6,1 l) in the slope define these temperatures, changes The although there is someambiguity, because the MFSvs. 1/T relationship is not exactly linear. The main results are listed in Table 2, where they are comparedwith the predictions of the empirical equations based on chemical composition reported by Barbosa et al.12) for T"*, and by Ouchi et al.13) for A,3. The distinct effect of strain rate on the T*, is evident, while the reheating temperature (1 100 or 1220'C for the Nbsteel, and 1220 or 1250'C for the high Ti-Nb steel) has no apparent influence. critical
recrystallization austenite-to-ferrite
Plate Rolling Simulation
In Fig. 1, typical
C 1995 ISIJ
0.03 0.03 0,03
hTi-Nb
cl'
constant at 2s~ 1, or in a few cases 0.2 s~ 1. The samples were continuously cooled at I'C/s, and the interpass time ti Wasfixed at 30 s. the schedule simulations, For the sheet rolling consisted of 6 roughing passes (strains of 0.3 for passes R1 to R4, O.4 for R5, and 0.5 for R6, separated by interpass times of 15s for R1 to R5), followed by 12 finishing passes, each of 0.24 strain separated by either 3 or I s interpass times. The numberof finishing passes in the torsion simulation was higher than in a sheet mill so as to reveal more clearly the trends in flow stress behaviour. A strain rate of 2s~1 was used in all the experiments. The temperature of the final roughing pass (R6) was adjusted to fall between I 100 and 975'C by varying the interval between R5 and R6 (60 to 200 s), and the temperature of the first finishing pass (F1) was sirnilarly adjusted to fall between I 050 and 900'C by varying the delay time between R6 and Fl (50 to 150 s). Cooling during finishing was achieved by opening the furnace after switching off the electrical power, and was at a rate of about 5to 6'C/s. The exit (F12) temperature varied accordingly, and could be as low as about 780'C. After finishing, the specimens were cooled at a rate of about 2 to 2.5'C/s into the vicinity of the A.3 tempera-
3.1.
0,041 0.041 O.050 0.034 0.03 0.034
~
1
I
Results
0.005 0,008 0.082 0.006 0.007
200
~'5
1
3.
N
Nb-Ti-Ni
The torsion tests were conducted on the computer controlled servo-hydraulic apparatus at McGill University in Canada. The machine, its construction, and previousdescribed operation been of have method ly. I '2,5, I o) For the plate rolling simulations, the simplified plate mill schedule often used at McGill was employed. The reheating temperature was 220'C, which corresponds to mill practice, except in somecases, where it was 100'C (for the Nb-Ti-Ni steel) or 250'C (for the high-Ti-Nb steel). The soaking time was 15 min. Fifteen or 18 passes were applied, consisting of constant equivalent strains of 0.30; the strain rate was held
.
Ni
240
plates.
ture
Al
Mo0.1
chemical compositions of the steels are given in Table l. Torsion test specimens (gauge length 20.0mm, gauge diameter 6.3 mm)were machined from pieces of rough rolled
v
Ti
l
(wt"/.)
steels.
ISIJ International,
Vol.
simulation Typical MFS vs. 1/T data for the finishing in Fig. with interruption times of ti shown = or s are 2a) for the Nbgrade, the behaviour of the other grades similar, as demonstrated for low-Nb being qualitatively and high-Ti-Nb steels in Fig. 2b). The figure displays the strong influence of ti on the flow stress behaviour. For
180
ti=30 s
Nbgrade
(plate)
160 140 Is
+
uJ oc
~ O JL L z
ri+
3s
l+
+
*
cl)
h c"
/
;:j:Fl~h/
,~
-~v)
/
120
/
ferritic
was formed
structure
during the sheet rolling simulations, and a dual grain size with tiny grains together with muchlarger ones was evident
several
in
as
cases,
shown
in
Fig.
3.
The
dependence of the ferrite grain size on the final pass (R15) temperature and the exit (F12) temperature is illustrated in Fig. 4. There is a general trend of decreasing grain size with decreasing temperature, although accompaniedby a fair amountof scatter. The ferrite grain sizes obtained are somewhatlarger than those reported by Samuelet al.1,2) and Bowdenet al.,7) especially if the lower exit temperature is taken into account. It should be noted, however, that the deformations of Bowdenet al.7) were applied isothermally while the cooling rate through the austenite-ferrite transformation was about in similarly O'C/s, cooling tests of Samuel continuous as 1
~-~~"~~ .
MFS=380000fr-225 .gOO'C
..
925'C
"'
/ // .
coarse, fully
7
..'
Lu ::
80
s
/
/ / ~~
Is
100
/
2
A relatively
MFS
3
ferrite
9
5
MFS I
was 3s.
grain size was determined metalloThe graphically by the linear intercept method for all the specimens. In the 15-pass plate rolling simulation, the final pass was applied near the A*3 temperature, and a fine, equiaxed ferrite relatively grain size, in the range to to ktm, resulted from the cooling rate of about 2.5'C/s. final
generally s, the MFS = I flattens
increases rapidly at first and then out, even though the temperature decreases by about to 6'C from pass to pass. For ti=3 s, there is a more gradual increase in MFS, especially at the highest entry temperature. In addition, the effect of entry (Fl) temperature is readily evident, with the increasing as the Fl temperature is reduced. in the Pass-to-pass "oscillations" can be seen in the Nb grade simulation with F1 OOO'C and ti = s. = This phenomenonwas most pronounced for the lowNb-Ti steel (Fig. 2b) but also visible for the other grades (except the high-Ti-Nb steel) whenthe Fl temperature
interpass time
Grain Size
Ferrite
3.3.
3 I
ti
I OOO'C)and the
was high (-
Sheet Rolling Simulation
3.2.
35 (1995), No. 12
,•~;~
/ ""950'C
.' g75'C . "I OOO'C
a)
60
(;~.
0.86
0.81
0.76
0.91
0.96
1OOO/T(K1) 180
:"cr_._,~
ti=3S
160
Fig.
Typical dual grain size after sheet rolling (FI =950'C. F12=798'C, t~ =3 s).
3.
,~
-=
hTi-Nb
140
cl) cl) LLJ
oplate
az
H 'l)
IZO
ti=
:~
O J LL
z
IS
15
E eF
ti=3S
100
INb-Ti
~
80
1oOo'C
lp'
5
CC
sml
•sml
~
,'
e
,,"' '
J,
,
e
~
ee
UJLL
60 0.76
0.81
0.86
0.91
Ref.[2]
0.96
1OOO/T(K~b 2.
~;
10 .''
~aC
b)
rl
"e
Z O UJ h
:~
rl
esheet
UJN
CO
LLJ
Fig.
simulation
o
simulations carried out on: a) Nb grade, the predicted onset of is markedby *, b) Iow-Nb-Ti and highprecipitation TiNb grades.
MFSvs.
l/Tcurves for the sheet
Ref.[7]
750
rolling
850
1050
950
EXITTEMPERATURE ('C)
Fig.
1525
4.
Ferrite
grain size as a function
of exit temperature.
C 1995 ISIJ
ISIJ International,
Vol.
the present tests were performed under continuous cooling conditions, and the cooling rate was about 2'C/s.
et al.1,2)
is
for pass R6 also promoted rapid recrystallization. austenite is in strain-free The precipitation of effect by sluggish, even though a kind of "conditioning" at higher previous deformation and recrystallization temperatures seems to accelerate subsequent precipitation at lower temperatures. 14'1 7) Generally, however, the o/o Precipitation in undeformed measured time for austenite is about 100s or longer.18) Thus, it can be
NbCN
5
precipitation occurred in the present tests before the start of finishing. (Even in precipitation deformed austenite, the onset of would require at least s of holding at 975'C under
concluded that no significant
Long Interpass Times The long interpass times, 30 s in
4.1.
the present tests, are rolling. plate reversing The slope considered to represent sharply Fig. in increases above plot of the vs. 1/T 920'C). (i.e. of about 0.84 below a temperature OOO/T=
l
As has been
is due to the transition from the occurrence of full softening between passes to the retention and accumulation of strain. This transition temperature is called the recrystallization temperature " T*.,6,il,12) or the "no
5
takes place during the period when o/o Precipitation place.14) methodfor predicting the T., or RSTbased recrystalon the interaction between precipitation and lization, has been described by Dutta and Sellarsl4) for isothermal conditions, and more recently by Bai et al. 15,16) under continuous cooling conditions. They have also analyzed the effect of processing variables on the T**. As shown in Table 2, the T** can be obtained
MFS
3
.
4.2.2.
MFS
This
retained from the sheet indicates that no is that the temperature for roughing stage. The reason selected to be above the T~*, and the pass R6 was minimumdelay between R6and Fl wasfairly long (50 s). The equation proposed by Dutta and Sellarsl4) for the
to.os
to,05,
in
Softening during Finishing
by Samuel et a/.1,2) and Pussegoda et al.3~5) was to employed. An equation was derived relating the MFS l/T at high temperatures under complete softening conditions (i,e. above the T~* during the plate rolling simulation). This was then used to calculate the multi-
Nb steels
factor for the stress at each pass. The resulting equation is shown in Fig. (MFS=3800001T-225), and typical stress-strain curves for all passes, "corrected" plication
2
x lO~ 20dge~4exp((2.75 x 105/T- 185) [Nb]) (1) x exp(300 OO0/8.31 T) ...................................
= 6.75
•
Here do is the original grain size in um, 8 is the strain, and [Nb] is the concentration of Nb in solution. The
C 1995 ISIJ
3
cooling simulation can be determined by correcting for the effect of declining temperature on the stress level in successive passes. For this purpose, the precedure used
is
time for 5o/o static recrystallization, can be used to demonstrate this:
A
Thedegree of softening betweenpasses in a continuous
values for the plate rolling simulations, once the the difference in pass strain is accounted for (plate
8=0.24).
2a).
MFS
Short Interpass Times 4.2. I MFSduring the Simulated Sheet Finishing Passes As shownin Fig. 2a), the MFSvalues for pass F1 of the sheet rolling simulations correspond quite well with
simulation
MFS
Iine structure (the MFS=3800001T-225 is is less especially than This true when F1 moderate increase in the slope of or equal to 950'C. the vs. l/T curve can also be detected after pass for the simulation where Fl =950'C and ti= s. F5 In order to understand these trends, the kinetics of softening and precipitation, and the possible interaction these between two processes, must be considered.
in Fig.
4.2.
sheet
I
recrystallized
low-Nb-Ti grade).
strain
below-which
the entry temperature. Moreover, it always remains above the level expected for a completely statically
approximately from Barbosa et al.'s experimental equation, although the calculated temperatures seem to be systematically higher than the measuredones (except for
8=0.30,
4.2.3
I
3
limit temperature" (RLT),14) and cor"recrystallization responds to the changeover from complete to incomplete static recrystallization during the delay tirne. Another is the "recrystallization stop temperature" used term o/o recrystallization taking (RST), corresponding to
simulation
Sec.
SRX
after roughing is completed long before meansthat precipitation begins.) Additional evidence that precipitation has not yet taken place can be seen in Fig. 2, where the flow stress levels during the earlier finishing passes substantially lower than at ti = s. This at ti = s aresignificant softening is taking place between suggests that these passes, which would not be the case if prior precipitation was interfering with recrystallization. Theoverall evolution of the flow stress during finishing depends strongly on the interpass time as well as on the Fl temperature. With ti = s, there is only a very slight after F2 (see Fig. 2a), even under increase in the conditions, continuous cooling down to 880'C or so. increases with pass numberand With ti = s, the decreasing temperature, the exact value depending on
this
A
conditions-see
isothermal
papers,1'2,6,11,12)
shown in several earlier
NbCN
II
I
5
(2) .................
This gives 0.45 s as the time for 950/0 recrystallization at 975'C, which is the lowest R6 temperature used. Equation (1) also showsthat the large strain of 0.5 used
considered in detail here.
MFS
t is
X=I -exp(-0.0513(t/to.05)2)
Three kinds of fiow stress behaviours can be in the present simulations, depending distinguished mainly on the interpass tirne. The first type is associated with long ti's (about 30 s), and the other two types with short ti's (~3 s). Similar behaviours were exhibited by the different grades, so that the conclusions that follow Only the Nb hold for all of them, at least qualitatively. grade
X, in time
recrystallized,
fraction
By contrast,
Discussion
4.
35 (1995), No. 12
1526
to a constant temperature 950'C, are shown in Fig. 5. Surprisingly, the resemble a single peak-type curve, which is commonly interpreted as resulting from DRX.I ~ 5,9,10)
ISIJ International,
150 Nb grade fl
~
~ ~
Il
/1
130
/f
l
110
F1=1000'C Tcor=950'C . /1
/1
I/
Il
ti=1
I If
,
c::
(1
995). No.
12
1 F1 =1 OOO'C, ti=3s
Oz
s
~
/
O c')
!r
0'5
J
~ Q
,
Ui
35
ti=3s
IlJ
Il/1
Vol.
OHm
~
90
measured
,
OH
o
70
predicted
d0=40um ,
o 1
23
a)
4 5 6 7 8 9 101112 PASSINTERVAL
50
2
3
2.5
4
3.5
1
5
4.5
STRAIN
Nbgrade
150
/1/1
f/
180
~
~UJ
F1 950'C
ti
Tcor 950'C
ti
r
t
/
/
f
l'
/~ ,
/1
110
a:
~
F1 = I OOO'C, ti=1
Oz
tO
3s I s --
predicted
CO
J
~
lllin l I /1
0.5
g
ll
OH
~
measured
90
O i 2 3 4 5 6 7 8 9 1011
12
PAss INTERVAL Fig.
Measuredand predicted softening
6.
with Fl
b)
50
2
25
3
3.5
4
4.5
Ii
F1 =950'O
~
b)
ti
for the simulations
= I s.
,
ti=3s
~ t
Stress vs. strain curves corrected to 950'C for the simulations with: a) F1 I OOO'C,b) Fl =950'C.
O co J 0'5
=
The amount of softening between passes was determined from these curves using the I o/, offset stress method.19) Each pass was always comparedto pass Fl The softening values measured for the Nb steel with Fl=1000 and 950'C are shown in Figs. 6 and 7 .
~
O_
measured
H o
~
predicted
o o 1
respectively.
These results reveal that there is little softening during the interval betweenthe first and second 100/.; and for passes, for Fl =950'C and ti=3s, ti s, X-6~/o. the However, of softening amount = increases up to 32 ~/o in the interval between F4 and F5 s. even for ti
I
= I OOO'C:a) =3 s,
i
5
STRAIN 5.
b)
o
70
Fig.
s
Z
X-
=I An important
point to note here is that there can be significant a amount of SRXduring sheet finishing at temperatures below the T~, determined in the plate rolling simulation, a point that has not always been clear in previous works.13'7) Alth6ugh there is at first insufficient time to allow a pronounced static recrystallization, the strain accumulates, as a result of which the SRXkinetics are accelerated (see Eq. (5b) below), the refinement of the grain size also acting to the samedirection. The work of Bai et a/.15) shows that at high strains under plate rolling conditions, T** be
as can 10was 850'C for Nbsteels. The measuredamountof softening can be compared with predictions from equations derived previously to describe the softening kinetics. Several of these have been
1527
23 45
a)
6 7 8 9
10 11 12
PAss INTERVAL
1 F1 =950'C
~
,
ti=1
s
~
O (D J
~ Q
0.5
- - - -
-)o~-
-
- - -
\
OH
\
measured
~
MRX -O-
predicted
b)
o O 1 2 3 4 5 6 7 8 9 10il
12
PAss INTERVAL Fig.
7.
Measuredand predicted softening for the simulations with Fl =950'C: a) Ii =3 s, b) Ii= s. I
a~.ter developed over the years for, statiQ recrystajlization a single pass under isotherrfral conditiops.14'19~22) The task becomes ~uite complex, howelyer, for multible
C 1995 ISIJ
ISIJ International,
Vol.
particularly
MRX in
995). No.
9)
and A=0.5
A=1 if X
2s- 28) This calculation
..........(3)
DRX
Nb
for
X;~0.1,
steels.
takes place after
by the open squares
employs the amountof softening that each pass, which depends in turn on
whether or not dynamic recrystallization during the pass. The critical given by9):
strain
8.=A •ep
e*
for the latter
Precipitation during Finishing In order to check whether softening is really retarded by strain-induced precipitation during finishing rolling, is appropriate it kinetics to evaluate the precipitation Dutta and under continuous cooling conditions. Sellarsl4) have proposed a model for the isothermal in deformed austenite containing precipitation of Nb. The time for o/o Precipitation, to.05p, is:
is
..........(4a) ....
where 8p
= 2.84 x
l0~4d0.5 e exp
)) 375 OOO
o.
17
8.31
•
NbCN 5
(4b)
'
T
to
and Ais a constant betweenabout 0.6 and 0.9. For C-Mn steels, A is commonlytaken to be about 0.8, but for the present Nb steels, a value for A of 0.65 was employed. If e* e., then static recrystallization occurs between are22): equations appropriate the passes, and XSRX =I to
exp(
-
50SRX=
-
O.693(t/t0.50
(~ 5.24 + 550[Nb])
(5a) " """ " SRX 10 ~ 188~4+ 77[Nb]d 2 '
x exp(330000/8.31 T) '
where dis the grain
..........(5b) ........
size at the start of a pass, and If e*>8*, in solution.
[Nb]
Nb
then during the pass in question, and the corresponding equations for softening immediately after unloading-that is, for MRX-are9): the concentration of dynamic recrystallization
is
XMRX = I -exp(
-
is
=
'
(6b) .
recrystallized grain size of a partially ith di, structure after the pass, was taken as equal to the dsRxi, recrystallized grain size as suggested for the uniform softening method,21,24,25) using the equations:
The average
C 1995 ISIJ
- 18- 1(~
'
exp(400 OO0/8.31 T))~0.5 •
where ~is the strain ratio for Nb. Park
2.5
x 1010
(8)
T3(In k)2
k
is the supersaturation rate, and et al.27) recently used isothermal
(PPT) data and conprecipitation-timeternperature (CCP) verted them into continuous-cooling-precipitation data using the additivity rule. The CCP curve so derived is located at slightly lower temperatures and longer times, as can be expected. Bai et al.15) adopted the same procedure to predict the T~* by considering the
study involved continuous cooling conditions in a sirnulation of seamless tube processing. In previous work, the complete accumulation of strain was assumedand introduced into Eq. (8). However, as shown above, the accumulation of strain may considerably accelerate softening in later passes, even when
1 T)
t0.50MRX 4.42 x 10~7~~0.59 exp(1 53 OO0/8.3
3x l0~6[Nb]
x exp(270 OO0/8.31•T) exp
steels.
(6a)
"""-"""-""
05p =
in interaction betweenprecipitation and recrystallization al.s) also used the plate rolling simulations. Pussegodaet rule to analyze the competition between additivity in Nb-V and Ti-V precipitation and recrystallization
initiated
0.693(t/t0.50~**))
in Fig. 7b).
4.2.3.
initiated
is
T
relatively high, roughing temperature of R6 = effect of the initial grain size is demonstrated 050'C. The 1 in Fig. 6a, showing an infiuence during the three first passes only. The overall trends of the softening predicted from the equations are seen to match rather well with the measured ones, in spite of numerousexperimental In the cases where constants used in the predictions. ti=3s, the softening was found to be due to SRX throughout. This was also true for the ti = I s simulation beginning at I OOO'C. However, for the ti = I s and Fl =950'C (Fig. 7b), the amount of SRXstays rather low during two first pass intervals which leads to the accumulation of strain up to high enough (8*=0.67; during pass F3, followed by 8*=0.52) to initiate metadynamicsoftening after the remaining passes (shown
..
if
•
shownin Figs. 6and 7for the Nbgrade. The initial grain Wastaken as 40 ,tm, which corresponds to the
final,
with
8.31
Examples of the results of the softening predictions based on these equations for F1 = I OOOand 950'C are
Decreasing temperatures can be taken into account, following Bai et al.,1 5,16) by using the additivity rule, or by adopting the concept of "temperature-compensated" time, as proposed by Sellars and Whiteman.23) However, interpass times were in the present case, the finishing allow the softening short be to enough judged to calculations for each pass to be performed at a single temperature corresponding to the average of the prior and subsequent passes. The accumulated strain for the ith pass, a.i, can be computedusing the uniform softening method:
+ ~(1 +Xi_ 1)e.=_ ,
)) ~ o 13 375 OOO .(7b)
size do
=ei
..........(7a) ....
( MRX dMRXi=1.37 x 103~~'exp
for
steels.9)
8.i
12
and
Nb
the case of
(1
for SRX,26) dsRxi=1.18.=-0.67A0.67 ui_1
where the temperature decreases continuously, and where there may only be partial recrystallization between passes. If there is only incomplete static recrystallization, as can happen in the case of short interpass times, then it is possible that enough strain will acto be initiated cumulate for dynamic recrystallization this would imduring a later pass. Upon unloading, mediately be followed by metadynamicrecrystallization kinetics are (MRX). Equations for describing
passes,
somewhatless well-known,
35
it
is
The
negligible
latter
after
the
first
finishing
pass. Clearly,
the
extent of softening must be evaluated and taken into account. In the present work, the effective strain and
1528
Vol.
ISIJ International,
1OOO
u 900
~l ~X~~;;1~~~~""
~~---J~
JO
H ~:
E H
DRX
- - _ _ _ J~]_ - __
:)
~
within 4s. In order to reach the critical strain for before that, the pass strain must be high (~:0.3) during the first few passes. The samepredictions can be applied to the other steels under investigation, if the kinetics of SRXand precipitation are known. The Dutta and Sellars model with modified constants has been employedfor V steelss) and Ti steels.32) Roughly speaking, the behaviour was identical in all the steels containing Nb, which suggests in the differences that there are only insignificant kinetics between Nb and softening and precipitation Nb-Ti steels, except the low-Nb-Ti and high-Ti-Nb grades. This is in agreement with the recrystallization
by precipitation
~_ ~
950
LLJ cx:
35 (1995). No. 12
'
~~~~~1 +1 ~~\ ~ e \.\ b \ '1 \ ~l\. , : ti=1
s
, '
\
,
850
' '
+ no accum.
UJ
800
CCP
;/+~/lL1~1
>K
Pred' accum.
X
fUll
accum.
~r~F ~l\
I
I
ti=3s
\1'T'
t
_ I~F
:
\ \.:~
t.T'
PTr Eq.(8)
Nbgrade
data reported for Nb and Nt)-Ti microalloyed steels29) and with the recent measurements for the present
lb
7SO O.
1
1
1OO
10
grades.30)
FINISHING TIME (s) Fig.
Cooling histories
8.
in
the
finishing
predicted onset of precipitation conditions of strain accumulation.
4.3. stage,
and the
under different
Ferrite Grain Sizes Producedby the Rolling Simulations
After the plate rolling simulations, the measuredvalues ferrite grain size were to ~m, in agreement with the values reported by Hodgsonand Gibbs22) for Nbmicroalloyed steels. If higher cooling rates are used, slightly finer grain sizes can be expected. For example, in previous laboratory roiling experiments, a cooling rate of 20 to 25'C/s applied to similar Nb- and Ti-microalloyed steels resulted in a ferrite grain size of about 5kLm.31) has been shown to be a more effective method for refining the austenite grain size than the pancaking of austenite below the T., temperature.1~5) Lowentry and exit temperatures are found to favour the formation of fine grain size.1~7) In the presen~,=~ests, softening seemedto tak,e place only when however, Fl ~ 950'C and ti ~ I s. The ferrite grain sizes measured are coarser than those reported by Samuel et al.1,2) despite the somewhatlower exit temperatures, and are also coarser than the grain sizes reported by Bowdenet al.7) after their isothermal tests. The difference between the present cooling rate (2 to 2.5'C/s) and that of the other tests (lO'C/s) can contribute to this difference. However, although a refinement of about I .2 ktm in grain size can be attributed to the higher cooling rate,32) this does not account for all of the difference. Dual grain sizes have frequently been observed (Fig. 3), supporting the view that partial SRX and pancaking of the austenite both occur in the final stages. This would then explain the lack of sufficient grain refinement. corrrelation between the steady state stress and the dynamically recrystallized austenite grain size was reported for a CMn-0.0280/,Nb and a C-Mnsteel by
7 9
of
fractional softening were calculated for each pass and for the subsequent pass interval using Eqs. (3) and (5), as described above, and the effective time for precipitation was evaluated, assuming complete additivity, in the
manneremployed by Pussegodaet a!.5) The predicted times to.osp and corresponding temperatures (CCP) are shown superimposed on the cooling histories for the Nb steel in Fig. 8. The 5'/* PTTcurve
DRX/MRX
also plotted for comparison, as predicted from Eq. (8) for a constant strain of 0.24 under isothermal conditions. Furthermore, the onset of precipitation under continuous cooling conditions of complete strain accumulation or in the absence of strain accumulation are indicated for certain conditions. The data are located at lower temperatures and longer times compared to the PTT starts curve. The prediction shows that the precipitation after F5 for F1 =1000'C, ti=3s, after F3 for Fl= 950'C, t~ = s and after F6 for Fl =950'C,*t. = s. Note that the test variables vary for the data, so that, if interpreted strictly, they do not form a single curve. The onset of precipitation is also marked by stars in Fig. 2a. For ti=3 s, a small change in the vs. llT curve can be detected after F4 for F1 =950'C (after F6 for F1 = OOO'C), i.e. one pass later than the predicted start of precipitation, which supports the assumption that precipitation, although to slightly higher extent than olo , is the reason for the diminished softening during the is
CCP
3
CCP
I
CCP
MFS
I
5
subsequent passes. This lcads to a more rapid increase
MFS 9
several passes
follow,
7
as did the to passes in the present tests. According to the previous calculation, precipitation should also start in the case of ti (after for for instance). Noincrease 950'C, F1 F6 = s = in could be detected after that, however. It can be suggested that if precipitation only occurs during the last few passes, its effect is not significant. Dueto the short ti, finishing only goes on for 6s after the start of precipitation, and therefore only a small amount of precipitation can be expected. With the lower entry is not effective temperatures. Fl ~ 900'C and ti ~ s, SRX and strain accumulation therefore takes place, followed in
Ievel,
if
still
MRX
A
Bowdenet
al.7)
MFSand
ferrite
An analogous relation
between the final grain size has been reported for IF steels.8) As shown in Fig. 9, the seems to have a similar relationship with the final ferrite grain size in all of the microalloyed steels studied here. Furthermore, the data consist of grain sizes from both the plate and sheet rolling simulations. The data from Bowdenet al.7) and Samuel et al.2) are also included in this figure, and all the curves have similar shapes.
I MFS
I
MFS
,
1529
C 1995 IslJ
ISIJ International.
e 15
O
E
plate
o sheet
rl
sml.
rl
sml.
Vo[.
(5) Due to pronounced static softening after a few passes at the entry temperatures of OOOto 950'C with the interpass time of s, sufficient strain for the initiation
3
of dynamic recrystallization
(5)
eF
o
UJ ~! co
~
partial
o
•\
(, UJ
o
\
H ~a:
5
a:
o
o
oo o o Ref.[2]
150
200 STREss(MPa) MEANFLOW
Relationship between ferrite of the final pass.
Frg. 9.
5MFSr
\ Ref.[7]
o 1OO
' 2.1•10
\
grain size and the
I
With
which was somewhatcoarser than commonlyreported after dynamic recrystallization. A dual grain structure static recrystallization. partial frequently formed by was last flow the stress of The mean pass is related to the ferrite grain size for both the plate and sheet rolling
250
MFS
simulations independently of the steel composition.
Conclusions
5.
Torsion testing was used for Nb and Nb-Ti microalloyed steels to simulate plate and sheet rolling schedules with interpass times of 30 s and or s, respectively. An analysis of the stress-strain data and measurementsof grain sizes allow the following conclusions to be drawn: (1) transition to from static recrystallization no-recrystallization conditions as well as the austenite phase transformation cause changes in the slope of the meanflow stress vs. the inverse pass temperature curve. The no-recrystallization temperatures and phase transformation temperatures determined from those changes for the microalloy steels are in reasonable agreementwith the predicted values. (2) The mean flow stress level in sheet finishing simulation depends on the entry temperature (1 OOOto 900'C) and the interpass time. During the first few passes, it remains the lower, the longer the interpass time is, but during the last finishing passes the opposite may
Acknowledgement One of the authors (LPK) would like to thank the Academyof Finland for providing financial support during his sabattical leave.
3 I
A
REFERENCES l)
2) 3)
5)
153.
N. Pussegodaand J. J. Jonas: ISIJ Int., 31 (1991), 278. L. N. Pussegoda. P. D. Hodgsonand J. J. Jonas: Mater. L.
Technol., 6)
7) 8) 9)
S.
8(1992),
lO) ll)
l 2)
Jonas: Proc. Int. Symp. on Mathematical Modelling of Hot Rolling of Steel, ed. by S. Yue, Canadian Inst. Mining and Metallurgy, Montreal, (1990), 99. J. W. Bowden. F. H. Samuel and J. J. Jonas: Metall, Trans. A, 22A (1991), 2947. A, Najafi-Zadeh, S. Yueand J. J. Jonas: ISIJ Int., 32 (1992), 213. C. Roucoules, S. Yue and J. J. Jonas: Proc. Ist Int. Conf, on Modelling of Metal Rolling Processes, London, Inst, Met., 165.
C. Roucoules, P. D. Hodgson, S. Yue and J. J. Jonas: Metall, Trans. A, 25A (1994), 389. S. Yue and J. J. Jonas: Mater. Forum, (1990), 245. R. Barbosa, F. Boratto. S. Yue and J. J. Jonas: Processing, Steels, Pittsburgh, Microstructure and Properties of
4
HSLA
by A. J. DeArdo, Minerals, Mater. & Met. Soc. AIME, Warrendale, USA, (1988), 51. C. Ouchi, T. Saripei and I. Kohazu: Trans. Iron Stee/ Inst. Jpn., 22 (1982), 214. B Dutta and C. M. Sellars: Mater. Sci, Technol., 3(1987), 197. D. Q. Bai, S. Yue, W. P. Sun and J. J. Jonas: Meta!1. Trans. A, ed.
13)
tions.
of kinetics of strain induced NbCN recrystallization, strain under partial accumulation and continuous cooling shows that at the entry temperatures of I OOOto 900'C, the precipitation starts after 5-2 passes at the 3s interpass time leading to a smaller softening and faster increase in meanflow stress level in the subsequent passes. With a I s interpass time, precipitation starts after 6passes at the entry temperature of 950'C, but no increase in the meanfiow stress takes place, due to the short time available for the precipitaAnalysis
precipitation
l4) l 5)
1530
24A (1993),
2151.
l 6)
D. Q. Bai, S. Yueand J. J. Jonas: Proc. Ist Int. Conf, on Modelling of Metal Rolling Processes, London, Inst. Met., London, (1993),
l 7)
E. Valdes: Influence of RoughingDeformation on Strain-Induced Steels. Precipitation in Thesis, University of Sheffield,
l 80
.
HSLA
PhD
(1988).
18) l9)
20)
.
C 1995 ISIJ
Sci.
63.
J. J.
London, (1993),
kinetics Predictions of the static recrystallization under multipass deformation, strain accumulation, grain refinement and continuous cooling conditions and comparison with the measuredinterpass softening indicate that static recrystallization and recovery are the significant processes which able to control the softening in sheet finishing simulation even at temperatures below the no-recrystallization temperature for plate rolling condi(3)
tion
H. Samuel,
Yue. J. J. Jonas and B. A. Zbinden: ISIJ Int., 29 (1989), 878. F. H. Samuel, S. Yue, J. J. Jonas and K. R. Barnes: ISIJ Int., 30 (1990), 216. L. N. Pussegoda, S. Yue and J. J. Jonas: Metal/. Trans. A, 21A F.
(1990), 4)
be true.
(4)
cannot accumulate, and
recrystallization
static
refinement.
OO o
I
is a process for grain the interpass time of s and the entry begins temperature of 950'C, dynamic recrystallization during the third pass causing a metadynamicrecrystallization between the latter passes. (6) Straining below the no-recrystallization temperature in the plate rolling simulation resulted in a fine ferrite grain size. The sheet rolling simulation led to a size that depended on the exit temperature, and grain
o
10
z
35 (1995). No. 12
Jonas and I. Weiss: Mel. Sci., 13 (1979), 238. D. Hodgson, J. J. Jonas and S. Yue: Proc. Int. Conf. on Processing. Microstructure and Properties of Microalloyed and Other ModernHigh Strength LowAlloy Steels, Pittsburgh, ed. by A. J. DeArdo, AIME, Warrendale, USA, (1991), 41. A. Laasraoui and J. J. Jonas: Metall. Trans. A, 22A (1991), 151.
J. J.
P.
ISIJ International, 21)
22) 23) 24) 25) 26)
Vol.
C. Devadas, I.V. SamarasekeraandE. B.Hawbolt: Metal/. Trans. A, 22A (1991), 335. P. D. Hodgsonand R. K. Gibbs: ISIJ Int., 32 (1992), 1329. C. M. Sellars and J. A. Whiteman: Met. Sci., 13 (1979), 187. P. Choquet, B, de Lamberterie. Ch. Perdrix and H. Biausser: 4th Int. Steel Rolling Conf.. Deauville, IRSID, France, (1987), B5.
Choquet. A. LeBonand Ch. Perdrix: Int. Conf, on Strength (ICSMA7), Toronto, (1985), 1025. Int. Conf. Hot Working and Forming Processes, Sheffield, ed. by C. M. Sellars and G. J. Davies, Met.
28)
29) 30)
P.
of Metals and Alloys C. M. Sellars: Proc.
London, (1980), 3. H. Park, S. Yueand J.
Soc.,
27)
35
S.
l 64 1
Jonas: Metall.
Trans. A,
23A (1992),
.
1531
995), No. 12 R. Priestner, C.ZhouandX.Yu:Proc. Istlnt. Conf. on Modelling of Metal Rolling Processes. London, Inst. Met., London, (1993), 331. (1992), 368. T. Siwecki: ISIJlnt.,32 L. P. Karjalainen: Steels '95, Proc. Int. Conf, on Steels, ed. by G. Liu, H. Stuart, H. Zhang and C. Li, China Science Technology Press, Beijing, (1995), 179. L. P. Karjalainen. K. Oksman and P. Steen: Proc. Int. Symp, on Inc., Poestenkill, Physical Simulation, Delft, Duffers Scientific USA, (1992), 43.
HSLA
HSLA
&
31)
32) J.
(1
C. M.Sellars and J,H.Beynon:Proc. Int. Conf. on High Strength LowAlloy Steels, ed. by D. Dunneand T. Chandra, South Coast Printers, Australia, (1985), 142.
C 1995 ISIJ
