AN EXTENDED 316L STAINLESS STEEL MODEL SUITABLE FOR INDUSTRIAL ROLLING

M. F. Abbod*, M. Mahfouf♦, D.A. Linkens♦ and C.M. Sellars♥

IMMPETUS
Institute for Microstructural and Mechanical Process Engineering:
The University of Sheffield
*
School of Engineering and Design, Brunel University, West London, UK
♦
Department of Automatic Control and Systems Engineering, University of Sheffield, UK
♥
Department of Engineering Materials, University of Sheffield, UK
Email: maysam.abbod@brunel.ac.uk

Abstract: During hot rolling processes, the material under deformation undergoes
different deformation conditions, i.e. temperature changes and strain rates. One particular
variable is the change in strain rate which can vary from low to very high values in
industrial rolling. Usually empirical models are used for predicting the material
characteristics but they are only valid within constrained limits. In this work an extended
model has been developed to predict the stress/strain characteristics of 316L stainless
steel material under a wider range of deformation conditions, i.e. high strain rates
changes which do not follow the equation-of-states laws. Copyright © 2007 IFAC

Keywords: 316L stainless steel, dynamic recrystallisation, non-equation-of-states, strain

rate, plane strain compression test.

1. INTRODUCTION to obey this equation-of-state; the instantaneous flow

During industrial processing of metal in particular the
rolling process, there are variations in the strain rate
that can produce transients in the stress-strain
relationships and in the resulting microstructure.
These transients may have significant consequences
on the subsequent processing and the final product
properties. Such transient behaviour must be
incorporated into existing empirical models so they
can predict the material properties under different
deformation conditions.
The transient behaviour following varying strain rate
deformations in a warm worked stainless steel (316L)
has been studied in this work. Plane strain
compression (PSC) tests with constant and changing
strain rate profiles were performed at a deformation
temperature of 1025°C. Quenching of the specimens
immediately after the tests allowed the evolution of
the substructure during the strain rate changes to be
analysed.
A first approximation for a mechanical equation-of-
state relating flow stress to the instantaneous value of
Zener-Hollomon parameter (Z) has been derived
(Abbod et al, 2006). For constant and slow changing
strain rate test conditions, the stainless steel is found
stress is determined by the instantaneous value of Z.
Following faster rate changes in strain rate there is
deviation from the state equation, and the
instantaneous flow stress during the strain rate change
is smaller than predicted. The deviation increases
with faster rate of strain rate change and a short
additional strain interval is required for the flow stress
to reach the expected level for the final strain rate.
In this paper, a constitutive equation model has been
developed for the prediction of the flow stress based
on the deformation conditions, namely the strain,
strain rate and the instantaneous temperature. This
work aims to investigate the effects of controlled
changes in strain rate on the instantaneous flow stress
of commercial purity austenitic stainless steel (316L)
under PSC testing.
2. DYNAMIC RECRYSTALLISATION
During deformation, work is carried out on the
material which implies the storage of dislocations in
the original structure. At the same time as work-
hardening increases the dislocation density, the
softening process of dynamic recovery takes place,
diminishing the dislocation density (the more
dislocations there are, the faster the recovery rate)
(Sellars, 1986). Dynamic recovery is a thermally
activated process and strain rate dependent: when the
temperature rises, the time necessary for this process
to occur is shorter. This is the reason why the stress-
strain curve is very dependent on the strain rate at
high temperatures. The work hardening may be
athermal, but dynamic recovery occurs from the
beginning of the curve, and this explains why the first
part of the stress-strain curve is temperature
dependent. This is schematically shown in Fig. 1,
where the microstructural evolution is illustrated
together with the corresponding stress-strain curves.
were carried out on a computer controlled
servohydraulic machine. The specimens were heated
to 1025°C and held for 60 seconds. They were then
transferred to the test furnace at 1025°C and then
deformed to the desired strains using constant or
variable strain rates in the range 0.5, 5.0 and 50 s-1.
The specimens were water quenched immediately
after deformation.
The PSC tests were utilised to obtain equivalent true
stress/equivalent true strain curves for different test
conditions as illustrated in Table 2 cases 1-3.
Furthermore, previous researchers’ data (Barbosa,
1985; Tanaka, 1987) (Table 2) were utilised to
develop the model based on calculating the stresses at
a strain of 0.1, stresses at 0.1 strain (σ0.1), and steady
state stresses (σss). These stresses were read directly
from the curves and were plotted against ln ε& (actual
value for the strain of interest) for each point of strain
and instantaneous temperature of deformation in
order to interpolate the strain rates for constant
stresses. The instantaneous temperatures during
deformation were calculated using a finite difference
package based on the dimensions of the samples and
the deformation conditions (Hand et al, 2000). The
temperature rise during deformation is shown in Fig.
2 for the different deformation conditions.

Fig 1. Schematic evolution of microstructural changes Table 2: Deformation conditions of the PSC tests.
during deformation.
case Strain Strain Temperature Author
3. 316L STAINLESS STEEL ALLOY MODELLING rate (˚C)
(1/s)
In this study, an austenitic 316L stainless steel was 1 1.0 0.5 1025
selected for the investigation of the hot deformation 2 1.0 5.0 1025 Abbod
behaviour of a low stacking fault energy FCC metal. 3 1.0 50.0 1025 (2006)
The choice of 316L stainless was motivated by the 4 1.4 5.24 950
aim of gaining a better understanding of the 5 1.4 0.05 1025
occurrence of dynamic recrystallisation in austenitic 6 1.4 0.5 1025 Barbosa
alloys. Furthermore, austenitic stainless steel alloys 7 1.4 5.0 1025 (1992)
have relatively low stacking fault energy and have 8 1.4 6.78 1100
been investigated extensively for more than thirty 9 1.4 0.05 950
years (Barraclough, 1974; Barraclough and Sellars, 10 1.4 0.5 950 Tanaka
1979; Ryan and McQueen, 1990;, Hughes et al 1974). 11 1.4 5.0 950 (1987)

The chemical composition of the 316L stainless steel

used in this investigation is given in Table 1. The 80
commercial composition material has a stabilized 70
microstructure with a grain size of about 100µm.
60
temperature rise (C)

950C, 5.0/s
Table 1 Composition of the 316L stainless steel used 50 950C, 0.5/s

in the present work. 40

950C, 0.05/s
1025C, 50/s
1025C, 5.0/s
30
Cr Ni C Mn Si Mo 1025C, 0.5/s
1025C, 0.05/s
wt% 16.7 12.2 0.024 1.5 0.29 2.63 20

10
Plane strain compression (PSC) specimens with 0
dimensions 50x30x10 mm were machined from the 0 0.5 1 1.5
as-received material so that test deformation is strain

parallel to the prior rolling direction. All specimens Fig 2. Temperature rise during deformation.
were coated with a thin film of DAG-2626 lubricant,
which was allowed to dry before testing. PSC tests
Over the whole stress range, the following general Table 3: Deformation conditions constant Zener-
equation was used: Hollomon tests.
sinh −1 ( Z / A)(1 / n )
σ= (1) case Strain Strain rate Temperature Z (1/s)
α
Values for α, n and A were obtained from the (1/s) (˚C)
stress/strain curves, and can be summarized by the 1 1.2 0.5 1025 1.74E+18
following equations: 2 1.2 5 1025 1.74E+19
3 1.2 50 1025 1.74E+20
Q = 460kJ / mol
εp = 0.473d00.02863Z−0.0353 (2) The constant Zener-Hollomon conditions are
achieved by changing the strain rate in order to keep
ε ss (isothermal) = 0.4691 Z 0.0241 (3) Z constant due to the temperature rise, Experimental
−1 (1 / 12.98476) results showed that there is no evidence of
sinh ( Z / 1.0E17)
σ0 = (4) mechanical transients during deformation for any of
0.015218 the rates of change in strain rate used. In all cases,
sinh −1 ( Z / 1.0E17) (1 / 1.737494) the flow stress appears to have reached the level
σ 0.1 = (5)
0.016068 expected for constant strain rate conditions by the end
−1 of the test. Barraclough and Sellars (1979) suggested
sinh ( Z / 1.0E17)(1 / 4.844028)
σe = (6) that there would be no cumulative effect on flow
0.008079 stress from a strain rate change, providing the rate of
sinh −1 ( Z / 1.0E17)(1 / 4.763623) change was less than some limiting value, Hlim, given
σss = (7)
0.00966 by:
where Q is the activation energy, εp is the peak strain,
εss is the steady state strain, σ0 is the initial stress, σ0.1 ∆ log Z  d log Z 
H lim = =  (10)
is the stress at 0.1 strain, σe is the non- ∆ε  dε  lim
recrysrallisation stress, σss is the steady state stress.
Below this rate, flow stress would depend only on the
The final equivalent stress is calculated using Eq. (8) instantaneous strain rate, and a mechanical equation-
for low strain while the effect of the softening process of-state would apply. No transient is evident for any
is expressed by Eq. (9) for high strain. of the rates of change used, 0.5, 0.34 and 0.055 for
cases 1-3 respectively, suggesting a very low
σ′ = 1.0{σ0 + (σe − σ0 )[1 − exp(−C′ε)]0.5} experimental value of H below the Hlim for austenitic
(ε < 0.7ε p ) (8) stainless steels.

     
1.4 However, it would be necessary to plot the expected
  ε − 0 .7 ε p
′ 
σ = σ − 1.0(σe − σss ) 1 − exp − 0.49 ( ) 
 flow stress paths for changing strain rates, assuming
  εss (isothermal) − ε p   
      an equation-of-state is obeyed, and compare these

(ε ≥ 0.7ε p ) (9) with the experimental results to confirm that there are
no transients. This requires the derivation of an
2
 σ − σ0  equation for the material and deformation conditions
where C ′ =  0.1 
 used, as developed in the next section.
 σe − σ0 
Fig. 3 shows the experimental and predicted flow
stress as a function of strain for the defined 4. EXTENDED 316L STAINLESS STEEL ALLOY
deformation conditions. Model validation was carried MODEL
out for constant Zener-Hollomon deformation
conditions given in Table 3. A non-equation-of-state behaviour of the material is
expected for high values of Hlim which is above the
300 limit of 4. Experimental work is being undertaken to
impose different strain rate changes during PSC tests.
250
The 8 cases are designed to have different changes in
200
the strain rate from low to mid to high level of strain
rates at different rate of changes. The first batch of
stress (MPa

1025C, 0.5/s
150 1025C, 5/s cases is selected to have a slow rate to the change in
1025C, 50/s
the strain rate. The test will start to a constant strain
100
rate up to a stain of 0.5, then the strain rate will either
50 decreased or increased to a different level within a
period of 0.5 strain. After the last period, the strain
0
0 0.2 0.4 0.6 0.8 1
rate will be kept constant for another 0.2 strain
strain period. This schedule should not exceed the Hlim
Fig 3. Experimental and modelled data for PSC tests. which means that the material behaviour should
follow the equation of estate (cases 4-7). The next
batch is designed to impose a higher rate to the 100
change in the strain rate. The test will start with
constant strain rate up to the strain of 0.9, then it will
change rapidly to the next level, either increasing or SR=0.5
decreasing the strain rate, within the period of 0.1 10
SR=5

strain rate (/
strain. The next stage is to fix the strain rate at the SR=50

level it reached for a period of 0.2 strain (cases 8-11). H=-2

H=-10
H=4
1
Three levels were selected for the constant strain rate, 0 0.2 0.4 0.6 0.8 1 1.2 1.4 H=10
0.5, 5 and 50/s with the strain rate changing from one
level to another, either increasing or decreasing.
Table 4 tabulates the cases with the specifications for 0.1
each. In the table, the Hlim for the first 4 cases is 2, strain
while for the next 4 cases it is 10. The former batch is Fig 5. Fast and slow strain rate change profiles.
well within the limits, while the latter is above the
limits (Prentice, 2001). Fig. 6 shows the stress/strain curves for the
experimental test of case 4 where the strain rate starts
Fig. 4 and 5 shows the strain rate profile for the at 0.5/s until a strain of 0.5, then ramping the strain
specified cases which are changing from 5 to 0.5/s rate to 5/s within a period of 0.5 strain. The strain rate
and vice versa, from 5 to 50 and vice versa. For the is kept constant at 5/s for a period of 0.2 strain. The
first batch the change is made at 0.5 strain, while for developed empirical model (Eqs. 1-9) was used to
the second batch the change is made at 0.9 strain. In predict the material behaviour which has given
both cases the strain rate is switched back to being similar results to the experimental curves. This has
constant at 1.0 strain and kept there for 0.2 strain. verified that within a Hlim of 2 the material follow the
equation-of-state.
Table 4: Deformation conditions constant Zener-
Hollomon tests. 200

case ε& Strain Temperature Hlim= 175

change rate (˚C) ∆log(Z)/ 150

@ (1/s) ∆ε
stress (MP

experimental
125
Strain empirical

4 0.5 0.5→5 1025 2 100

5 0.5 50→5 1025 -2

75
6 0.5 5→50 1025 2
7 0.5 5→0.5 1025 -2 50

8 0.9 0.5→5 1025 10 0 0.2 0.4 0.6

strain
0.8 1 1.2

9 0.9 50→5 1025 -10 Fig 6. Experimental and modelled data for PSC: case
10 0.9 5→50 1025 10 4.
11 0.9 5→0.5 1025 -10
Fig. 7 shows the stress/strain results for case 10 where
All the cases were experimentally performed using the test starts with a constant strain rate of 5/s until
the Servotest machine which can be programmed to 0.9 strain. Then the strain rate was ramped rapidly to
perform a specific strain rate profile. 50 within a period of 0.1 strain which gives Hlim = 10.
At strain 1.0, the strain rate was kept constant at 50/s
100 for a period of 0.2 strain. It can be seen that the
empirical model has “lead” the experimental results
which means that for such a case the material
10
SR=0.5 behaviour does not follow an equation-of-state model.
SR=5
This is due to the fact that the Zener-Hollomon
strain rat

SR=50
H=-2 parameter is not constant whereas it is assumed to be
H=-10 constant for fixed slow varying strain rates.
H=2
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 H=10
In order to include the effect of the rapid change in
the strain rate, an extension to the model has to be
0.1 included which is in the form of a factor, namely
strain Z_factor, which is a function of d log(Z)/dε. The new
Fig 4. Fast and slow strain rate change profiles. term is given in Eq. (11) which has two parameters to
adjust, scale and the power m. The two parameters
should be optimised to include all the test conditions.

Z _ factor = scale × ( d log( Z / Z 0 ) / dε ) m (11)

 conditions, which are characterised by sharp changes
250 in the strain rate, the empirical equations do not
predict the material behaviour accurately. The model
225
was further extended using experimental work to
conditions where the strain rate changes are high.
Such conditions are usually experienced in the rolling
stress (MPa

experimental
200
empirical process where the strain rate changes can vary from
small to very large figures. The developed extended
175
model can accurately predict the material behaviour
for high strain rate changes.
150
0.5 0.6 0.7 0.8 0.9 1 1.1
strain

Fig 7. Experimental and modelled data for PSC tests: 6. ACKNOWLEDGMENTS

case 10.
The authors gratefully acknowledge the UK EPSRC
The extension is added to Eq. (8). The new formula is (Engineering and Physical Sciences Research
shown in Eq. (12). Council) for their financial support under grant
number GR/R70514/01.
σ′ = (1 + Z _ factor) × {σ0 + (σe − σ0 )[1 − exp(−C′ε)]0.5 }
(ε < 0.7ε p ) (12) 7. REFERENCES

The Z_factor term was added to the equation with an
offset of 1 which is for the purpose of eliminating its
effect for the normal conditions (Z_factor = 0).
For all the four cases (8-11), the parameters scale and
m where optimised to get the best fit for all the cases.
The optimised parameters were as follows:
m = 0.1
scale = 0.02
Abbod M.F., Mahfouf M., Linkens D.A. and Sellars
C.M. (2006). Modelling of dynamic recrystallisation
of 316L stainless steel using systems approach.
THERMEC 2006, July 4-8, 2006. Vancouver,
Canada, Editor: T. Chandra, Publisher: Trans Tech
Publication Ltd, Switzerland.
Barbosa, R.A.N. (1985). Simulation of hot working of
austenitic stainless steels. Thesis (PhD), University of
Sheffield, Dept. of Metallurgy.

Fig. 8 shows the experimental, empirical model and
the extended model prediction for case 10. The
introduction of the model extension has made the
prediction more accurate by removing the “lead”
prediction.
240
Barbosa, R.A.N. and Sellars, C.M. (1992). Static
Recrystallisation of Type 316L Stainless Steel under
Hot Working Conditions, Recrystallisation'92, Proc.
Int. Conf. on Recrystallisation and Related
Phenomena, San Sebastian, Spain, 1992, eds. M.
Fuentes and J. Oil Sevillano, Trans Tech Publications,
Switzerland, pp. 461-466.

230 Barraclough, D.R. (1974). Hot working and

220
recrystallisation of a stainless and a low alloy steel.
Ph.D. thesis, University of Sheffield, UK.
stress (MPa

expt
210 hyper
non-equ
200
Barraclough, D.R. and Sellars, C.M. (1979). Static
recrystallisation and restoration after hot deformation
190
of type 304 stainless steel. Metal Science, Vol. 13, pp.
180 257-267.
0.9 0.95 1 1.05 1.1
strain
Hand, R.J., Foster, S.R. and Sellars, C.M. (2000).
Fig 8. Extended model prediction for case 10. Temperature Changes during Hot Plane Strain
Compression Testing. Mat. Sci. and Tech. Vol 16, pp.
442-450.
5. CONCLUSIONS
Hughes, K.E., Nair, K.D. and Sellars, C.M. (1974).
A dynamic recovery constitutive model has been Temperature and Flow Stress during the Hot
developed incorporating experimental testing Extrusion of Steel. Metals Tech., pp. 161-169.
conditions and validated on constant Zener-Hollomon
deformation conditions. The empirical model can Prentice, C. (2001). Effect of Strain Rate History on
predict accurately the material properties for limited Warm Deformation of Interstitial Free (IF) Steel.
conditions. Such conditions are limited by the Thesis (PhD), University of Sheffield, Dept. of
changes in the strain rate which is denoted by Metallurgy.
log(Z)/log(ε). For high extreme deformation
Ryan, N.D. and McQueen, H.J. (1990). Comparison
of dynamic softening of 301, 304, 316 and 317
stainless steels. High Temperature Technology, Vol.
8, No. 3, pp. 185-200.

Sellars, C.M. (1986). Modelling of structural

evolution during hot working processes, Proc. RISØ
international symposium on annealing processes:
recovery, recrystallisation and grain growth, pp. 167-
187.

Tanaka, A. (1987). Effect of non-isothermal

multipass hot deformation on strength and
recrystallisation kinetics, Thesis (MPhil), University
of Sheffield, Dept. of Metallurgy.