Detailed Technical Report by TU BERGAKADEMIE Freiberg A
Author  M Schmidtchen  R Kawalla 
Pages  305352 
Page 305
In the interpretation of industrial forming processes the coupling of single process steps plays an increasingly more important role. This applies particularly to continuous forming technologies, like the hot rolled strip and wire production or the open die forging. Since the influencing mechanical properties are increasingly carried out via the thermo mechanical treatment, the specific development of the microstructures gains a growing meaning.
The microstructure is for its part influenced by the tension and strain condition in the forming step as well as the interpass times, the material composition and the temperature regime.
The material transport during the forming passes is inhomogeneous over the cross section. A homogeneous initial state therefore results after the first passes in inhomogeneous strain distributions and therefore in an inhomogeneous change of the microstructure itself (locally different recrystallisation, grain growth and precipitation). These changes in microstructure influence the local material transport and the forming behaviour in the following passes now. In the common forming processes an additional stress and strain reversal will take place. Here a situation occurs which common simulation methods (empirical, numerical) can handle only insufficiently. The origin of this situation can be seen in the constitutive models used till now.
The aim of the partial projects for the individual partners in this ECSC multipartner project can be deduced from this point of view for (Fig. A5.1): A significant improvement on the simulation possibilities shall be found by improving the constitutive models.
Therefore rolling tests were carried out as benchmark for the simulation, material tests with different materials and test procedure as well as corresponding modellings in the individual partial projects. The material tests were carried out with different materials to guarantee a correct modelling of different softening behaviour. Therefore also different test procedures were used. The influence of different stress and strain states should be included here. By modelling of the results of the material tests and the rolling trials the quality of the received constitutive equations should be checked.
The main task of the TU BAF in this multipartner project consisted in delivering the experimental results from the rolling tests (geometry, rolling forces, torque and microstructure). An additional modelling of the strain hardening behaviour according to the Freiberg approach and its application in FE computations of single tests was the second topic of the partial program.
In detail the partial program of TU BA Freiberg was:
As part of the objectives set up for the benchmark exercise in this multipartners ECSC project (Fig. A5.1), TU BERGAKADEMIE Freiberg has carried out experimental studies of the development of the austenitic grain size in high and low speed rod rolling mill withPage 306involvement of other partners as well as low speed forward and reversal plate rolling (Fig. A5.2).
The experimental work of TU BERGAKADEMIE Freiberg is to be subdivided into three main phases:

Experimental work on rod rolling of medium CMnsteel with high and low rolling speed and on plate rolling of medium CMnsteel.

Investigation of the stress – strain relation of the medium CMnsteel used.

Visioplasticity investigations of the material flow in the roll gap during plate rolling.

Microstructural investigations of initial and final grain size.
The experimental results obtained in the rolling tests will be used for a benchmark test of existing numerical models with strain reversal. In detail three steps were done:
modelling of the stress – strain behaviour of C – Mn Steel by use of the Freiberg model
Computation of the rolling trials done in the experimental part (empirical model for rod and plate rolling; FE  computation)
Investigation of the softening behaviour at higher strain rate
Computation with softening behaviour
For all trials the continuous high speed rolling mill at TU BA Freiberg in rod configuration was used (see Fig. A5.3 – Fig. A5.5). All data for high and low speed rolling are to be seen in [A5.1].
The roll pass design (roundoval; 10 mm >6 mm) were held constant during all trials except trials at low speed using F1 and F2 at different strain. Geometry and rolling speed for all stands are presented in [A5.1] and Fig. A5.4.
At high speed (32 m/s) a configuration as to be seen in Fig. A5.3 was used (without water basin). At low speed (5 m/s) the final cooling was performed inside the turbulent cooling line. While the rolling speed was held constant (at 32 m/s or 5 m/s respectively) the heating temperature, heating time and entrance temperature at F1 was varied.
Temperature measurements were made during conductive heating, before entrance into stands F1  F4 and at the laying head. In all stands the rolling force and rolling torque were measured.
Page 307
Due to a fast increase of oxide scale and heat lost at the surface the measurement of surface temperatures was compared with data from thermocouple in the bulge at the same position. All measurements were done using the conductive heating of the continuous rolling mill.
The difference of the measured temperature increased. A maximum value of 140°C was obtained after heating for 10 min. This data were used to keep the temperature nearly constant during heating.
To optimise the cooling rate two series of tests were performed using a mixture of water and ice respectively water only. Water and ice was preferred due to thick layers of scale after long heating times. For short heating times (below 5 min) cold water cooling only is also useful.
Owing to 0.2% S a lot of segregation appeared. Therefore the austenitic grain size differs over the cross section. Examples of the obtained austenitic grain size are given in Fig. A5.12.
For plate rolling trials a two high reversal rolling mill of TU BAF with roll diameter of 360 mm was used. In Fig. A5.6 the rolling technology is given. The specimens were reheated in a heating furnace. Before descaling and before entrance into the roll gap the surface temperature was measured. This was also done immediately after rolling. During rolling torque and rolling forces were measured. With the first temperature signal the time was measured. The interpass time could be deduced from the rolling force – time signal. After rolling the specimens were cooled immediately in a turbulent water bath.
For all trials the continuous high speed rolling mill at TU BA Freiberg in rod configuration was used (see Fig. A5.3 and Fig. A5.5). All data for high and low speed rolling are given in Table A5.2 –A5.6 of [A5.1].
The roll pass design (roundoval; 10 mm >6 mm) was held constant during all trials except trials at low speed using F1 and F2 at different strain.
At high speed (32 m/s) a configuration as to be seen in Fig. A5.3 was used (without water basin). At low speed (5 m/s) the final cooling was performed inside the turbulent cooling line. While the rolling speed was held constant (at 32 m/s or 5 m/s respectively) the heating temperature, heating time and entrance temperature at F1 was varied.
Temperature measurements were made during conductive heating, before entrance into stands F1  F4 and at the laying head. In all stands the rolling force and rolling torque were measured.
High speed rolling was performed using all stands F1F4, water cooling and laying head. The reheating temperature was 1200°C and 1150°C respectively. The time for heating was also varied (approx. 10 s, 60 s; see also [A5.1] Table A5.3 – A5.6).
Page 308
The length of the rod reheated was 5 m (insulating bricks/air). On both sides the cold parts outside the insulating bricks were cut. After cooling in air for:
1200°C > 1100°C: ~10 s
1200°C > 850°C: ~25 s
1150°C > 1100°C: ~ 3 s
1150°C > 850°C: ~25 s
The temperature for rolling was reached.
Rolling at low speed was performed using
all stands F1  F4 and water cooling,
Stand F1 – F2 and water bath and
Stand F1 only and water bath.
The length of the rod reheated between insulating bricks was 5 m. For rolling trials using all stands F1F4 the reheated rod was completely rolled whereas during the trials with F1F2 and F1 only the first part of the rod (approx. 1 m) was used.
Due to a low speed in the last stand F4 the rolled rods stopped inside the turbulent cooling line (trials using F1 – F4).
During all trials using F1F2 or F1 only the rolled rod had to pushed into stand F2 and into the water bath manually. Thus a break of about 2 s exists between each pass.
The reheating temperature was as similar as in high speed rolling, 1200°C and 1150°C respectively. Also the time for heating was varied (approx. 10 s, 60; see also Table A5.3 and A5.6. of [A5.1]).
The initial grain size determinates significantly recrystallisation processes [A5.1]. Due to technical restrictions only reheating times of 10 s and max. 60 s at 1200°C could be realised. Thus a striven austenitic grain size of 200 µm wasn’t reachable.
The microstructure of the steel grade used was inhomogeneous due to precipitates and segregation's. Thus, inhomogeneous recrystallisation will be expected.
For the given rolling speed and a fixed distance between the stands the interpass time is also given. To obtain different effects of the recrystallisation on the rolling forces the time for static and dynamic recrystallisation must be changed.
From Experimental data (in [A5.1]) obtained at Swinden Technology Centre (STC) the dependence of the recrystallisation time on temperature, initial grain size and strain is known.
The lower the rolling temperature and the initial grain size the lower the time for full recrystallisation. Increasing the strain will give the same results.
Page 309
Whereas in [A5.1] an initial grain size of 100 µm was used to obtain t0.95 ~ 0.006 s (1100°C) and t0.95 ~ 0.24 s (850°C) the initial grain size for the rolling trials is in the range of 50 µm depending on the reheating parameter. Thus only for rolling trials at high rolling speed and low temperature unrecrystallised structures will be obtained. Variations in thermal conditions and rolling speed (interpass time) was used to affect the recrystallisation process during rolling (full and particular recrystallisation).
In the rolling trial two different sets for the rolling speed were used. The related sets of interpass time are given in Table A5.1. In [A5.1] the variations of the initial thermal conditions are reported.
During the rod rolling trials parameter were varied to study the effect of:
Reheating temperature,
Reheating time (at const. Temperature),
Rolling speed,
Temperature at the entrance of F1.
Strain variations were done due to the dependence of spreading on the rolling speed and temperature (const. roll gap geometry see [A5.1]).
From a plot of the rolling forces v time it is known that a stationary state was reached. Dependent on rolling speed and rolling temperature three principal trends of rolling forces were found (trend 13 in Fig. A5.7; A5.8). But all trials plotted in Fig. A5.7 results in similar final grain sizes (Table A5.2).
In general it is to be seen in the rolling force plots that the rolling speed and average temperature affects the recrystallisation process in a complex manner. But microstructural investigations of the material after the last pass (F4) had shown no significant differences in the final average austenitic grain size. Thus more detailed investigations of the microstructure after F1 and F2 are necessary to quantify the influence of the parameter given above.
The qualitative effect of reheating temperature and time can be derived from the development of the rolling forces during the whole rolling process (Fig. A5.9):
The trend curves in Fig. A5.9 represent the principal curves 1 (rolling speed 5 m/s) and 2 (rolling speed 30 m/s) in Fig. A5.7 and A5.8.
A higher reheating temperature results in lower rolling forces for constant rolling speed.
All trials plotted in Fig. A5.9 started at 850°C in the first pass F1. For different reheating temperatures similar cooling times (26 s … 28 s) were necessary to reach 850°C. Due to different reheating temperatures grain growth reaches different stages, for specimen reheated up to 1200°C a more homogeneous microstructure but also more scale occur. For the freecutting steel used the scale forces lubrication during rolling. So finally the reason for a lower rolling force can’t be attributed only to grain growth. Therefore a detailed analysis of the microstructure obtained after the first two passes are necessary. Computation of a single pass and comparison with experimental results will also give some information about the significance of friction and recrystallisation.
Page 310
During the variation of the rolling speed the ration of the speed between the stands was held constant. The effect of speed variations is to be seen in Fig. A5.9 and A5.11.
At constant thermal initial conditions (Fig. A5.11) different trends in the rolling forces were obtained. Therefore only particular recrystallisation can occur. The trends in temperature (at F2F4) for the trials given in Fig. A5.11 are similar but different in its absolute value. A possible reason is the shortening of the interpass time with increasing the rolling speed (Table A5.1).
At constant rolling speed the temperature at the entrance will also affect the trend of the rolling forces (Fig. A5.10). The rolling forces not only increase with higher temperatures. The trends differs also in the inclination inbetween the passes. Similar results available at lower speed (see Fig. A5.7). These behaviour can be attributed to the material behaviour in the transformation range.
While differences in rolling force are visible, the final grain size (behind F4) will be only slightly affected by all parameter changes. Smaller grain sizes can be measured for higher reheating temperatures or higher reheating times and higher rolling speed at lower temperatures in the entrance of F1.
More detailed studies of the development of the microstructure and mechanical values especially of the first two passes and comparison with computational results will help to get more information.
Due to a slow evolution of the grain size in CMnS steel grades during the reheating trials additional investigations of the grain growth behaviour were done. The final results are: no significant grain size evolution below 1200°C will occur. This is consistent with date from literature (Fig. A5. 13). CMnS steel grade tend to show abnormal grain size evolution.
For a detailed study dimensionless plots similar to the method given below for flat rolling is necessary. Therefore continuous and discontinuous stress strain curves must be predicted in upsetting tests.
Additional to the benchmark trials these pretest were performed:
Page 311
The CMnS – steel grade delivered by TU Freiberg with an initial diameter of 25 mm was prerolled in an open three high train at TU Freiberg. Thus a 10 mm x 6 m rod was prerolled within similar parameters.
The CMnNb – steel grade delivered by Corus with plates of 30 mm thickness was particular prerolled down to 6.7 mm within two passes. Thus two sets of specimen with different initial thickness were available to investigate the significance of the rolling geometry, to vary the rolling parameter and to improve the cooling conditions after the last pass. The geometry of the rolling specimen was fixed by the geometry of the turbulent water bath. Prerolling was performed at similar strain – time regime at a rolling temperature above 900°C.
Prerolled rods and plates were cooled down in air.
Temperature measurements were performed with pyrometer for rod rolling and plate rolling as well. During reheating or cooling parallel measurements with thermocouples were used for a calibration. Therefore the influence of oxide scale during temperature measurement could be reduced.
To investigate the evolution of the austenitic grain size of CMnNb steel grades a fast cooling is necessary to obtain a martensitic microstructure. From a database of TU Freiberg the transformation behaviour of the CMnNb steel grade was computed (Fig. A5.16). As can be seen in Fig. A5.16 a cooling time of t8,5 = 1.3 s is necessary to obtain martensite. Lower cooling rates will result in a bainitic or ferritic/pearlitic microstructure. These microstructure will affect the investigation of the austenitic grain size.
On the basis of these data computations of a simplified cooling behaviour with maximum heat transfer were undertaken to predict the maximum thickness (exit of the roll gap after the last pass) for the cooling rate necessary (tmax=1.3 s).
From Fig. A5.15 a cooling time t8/5 valid for the center of the plate is to be seen in dependence on the thickness for a given width and length.
For a maximum cooling time of 1.3 s a maximum thickness of 15 mm was computed. Real heat transfer conditions will lower the cooling rate. Therefore a smaller maximum thickness must be used in the rolling trials. With these geometric conditions cooling trials with plates of 15 mm and 7 mm thickness were performed (Figs. A5.17 and A5.18). Finally we obtain from the cooling tests: only specimen with a final thickness below 7 mm will cool fast enough to obtain martensitic microstructure.
The objective of the performed plate rolling trials was to investigate the connection between rolling parameter, RX kinetik and the evolution of the austenitic grain structure during forward or reversal rolling.
Page 312
Main objective of the trials were to perform a benchmark test and to show the effect of different strain and stress histories on rolling forces and microstructural evolution.
From all partners the following guidelines for the plate rolling trials were specified:
Material:
CMnNb steel (0.15% C, 1.4% Mn, 0.37% Si, 0.03% Nb, 0.017% Cr, 0.002% Mo, 0.02% Ni, 0.033% Al, 0.001% Ti, 0.003% V, 0.002% Sn, 0.007% Cu, 0.006% N, 0.016% P, 0.009% S)
Plate size: W120 mm, L400 mm, H30 mm
Mill facility: Freiberg Reversing Stand Mill, STC Plate Mill
Rolling speed: 1 m/s; Roll diameter: 360 mm (Freiberg);
Rolling Schedule:
Heating temperature: 1200°C (30 min) Rolling temperatures: 1000°C, 900°C, 850°C Rolling trials:
1st trials: Only 1 pass: 40% reduction (12 mm/30 mm=40%), water quench.
2nd trials: 2 passes with the first pass 40% reduction (12 mm/30 mm=40%) and the second pass in the same direction for a further 50% reduction (9 mm/18 mm=50%), water quench.
3rd trials: 2 passes with the first pass 40% reduction (12 mm/30 mm=40%) and the second pass in the reversal direction for a further 50% reduction (9 mm/18 mm=50%), water quench.
Measured values:
(a) Rolling force and torque
(b) Temperature at entry and exit
Microstructure studies:
(a) Grain size and microstructure photo BEFORE rolling (1000°C, 900°C, 850°C)
(b) Grain size and microstructure photo AFTER rolling (water quench)
Rolling trials with two different initial thicknesses and rolling geometry were realized. In general as it was discussed above the specimen delivered from Corus (W120 mm, L400 mm, H30 mm) had be cut into smaller pieces due to the geometry of the cooling bath and the maximum rolling force of the reverse rolling mill.
Page 313
A length of 120 mm (30 mm initial thickness) and 162180 mm (6.7 mm initial thickness) respectively were used. From Table A5.3 and A5.4 the initial width is given.
The different rolling conditions will result in different rolling parameter as discussed below. Focus was set on a similar ratio of contact length in rolling direction Ld and mean high of the roll gap hm. The rolling schedule was set as given in the guidelines above. Also rolling conditions at lower strain were used:
In this series of trials the rolling schedule given in the guidelines using smaller specimen (H30 mm x W80 mm x L120 mm) was realised. Due to the final thickness of the rolling specimen of 15 mm and 8 mm respectively no martensitic microstructure was obtainable (Fig. A5.19). At an equivalent strain of [VER SIMBOLO EN PDF ADJUNTO] per pass the rolling trials were performed in the region of dynamic recrystallisation. Decarbonisation could be observed in the surface region of the specimen (Fig. A5.19).
This series of rolling trials were performed using prerolled specimen with a reduced initial thickness of 6.7 mm (H6.7 mm x W100 mm x L162...180 mm). After the 1 pass the specimen (thickness 5 mm) were cooled at a higher cooling rate. Thus the t8/5 time in the center of the specimen is below 1.3 s and a martensitic microstructure is found.
The logarithmic strain [VER LOGARITMO EN PDF ADJUNTO] is used for computations. It is the total strain of a cross – section at the end of a pass and in its nature a mean value. In real processes the logarithmic strain is distributed inhomogeneously over the cross – section and the roll gap length. Thus for a calculation of the reference entities a mean value will be used.
To calculate the mean strain rate in the roll gap an equation of Hoff and Dahl
[VER FORMULA EN PDF ADJUNTO] was used. A good approximation within a range of [VER SIMBOLOS EN PDF ADJUNTO] can be assumed [A5.6].
The interpass time is given from rolling force v time measurements. At the beginning of the 1st pass the count starts and ends at the beginning of the 2nd pass. To obtain the real interpass time the time for one rolling pass have to be subtracted from the counted time. In our case this are 0.013 ... 0.05 s and can be neglected.
Page 314
Reheating was performed 30 s in air at 1200°C. A view seconds after reheating 1150°C were measured at the surface. This is due to scale growth and radiation. In the core of the specimen a temperature of 1200°C will last (A5.1).
The trials start at a higher temperature (rolling temperature + 50°C) before descaling. In front of the rolling mill the temperature at the entrance of the roll gap were measured. The pass starts when the temperature was exactly reached. At the exit of the roll gap temperature measurements were performed too.
After the first pass the rolled specimen were manually moved back to the entrance of the roll gap and for reversal rolling turned around 180°. This manual operation allows the interpass time to be minimised compared to a change of the rolling direction of the mill. On the other hand deviations in the interpass time were caused. Due to additional steps the interpass time for reversal rolling is longer compared to that of forward rolling.
To plan the rolling trials, to compute typical parameter and to study the transformation behaviour near A3 under forming conditions upsetting tests were performed. All specimen were reheated within 30 min at a constant temperature of 1200°C. Upsetting tests were done in a temperature range of 600°C up to 1200°C. To investigate the temperature range of phase transformation A3 during the forming process plots of the yield stress at a constant equivalent strain for selected equivalent strain rates and temperatures were computed (Figs. A5.22 and A5.23).
The phase transformation was obtained at a temperature range of 750°C800°C. To avoid phase transformation during rolling the trials must be performed at a temperature above 850°C.
The yield stress was calculated using the Freiberg model equation for the continuous stress – strain curves (m1…m8 constants for approximation):
[VER FORMULA EN PDF ADJUNTO]
The experimental data were fit to the model within a multiple approximation method (coefficients see Tables A5.5A5.6).
At TU Freiberg additional quasiadiabatic compression tests were performed to investigate the softening behaviour at higher strain rates. This value will be used to correct the dimensionless plot of the related forming resistance. Therefore compression tests at constant temperature (1000°C) were examined with different interpass times (0.5; 1; 5; 10; 50 s). The strain for the first step was 0.15 and 0.3 respectively. A strain rate of 1/s and 10/s was used.
Page 315
The softening fraction was computed using the 0.2% offset method and temperature corrected stress – strain curves. Results for the t0.5 parameter are to be seen in Figs. A5.31A5.33.
At similar strain, smaller strain rate and higher temperature (1100°C; 1150°C) results from torsion test have shown higher values for the t0.5 parameter [A5.3]. A similar dependence on strain and strain rate was predicted from torsion tests. Increasing prestrain strain rate and temperature will shorten the time for a certain value of fractional softening (Figs. A5.32 and A5.33). For increasing strain rate a fractional softening of 0.8 ... 1 was obtained for the interpass times of the plate rolling trials (5 ...10 s).
The initial grain size determinates significantly recrystallisation processes [A5.1]. Due to technical restrictions only reheating times of 10 s and max. 60 s at 1200°C could be realised. Thus a striven austenitic grain size of 200 µm wasn’t reachable.
The microstructure of the steel grade used was inhomogeneous due to precipitates and segregation's. Thus, inhomogeneous recrystallisation will be expected.
For the given rolling speed and a fixed distance between the stands the interpass time is also given. To obtain different effects of the recrystallisation on the rolling forces the time for static and dynamic recrystallisation must be changed.
From Experimental data (in [A5.1]) obtained at Swinden Technology Centre (STC) the dependence of the recrystallisation time on temperature, initial grain size and strain is known.
The lower the rolling temperature and the initial grain size the lower the time for full recrystallisation. Increasing the strain will give the same results.
Whereas in [A5.1] an initial grain size of 100 µm was used to obtain t0.95 ~ 0.006 s (1100°C) and t0.95 ~ 0.24 s (850°C) the initial grain size for the rolling trials is in the range of 50 µm depending on the reheating parameter. Thus only for rolling trials at high rolling speed and low temperature unrecrystallised structures will be obtained. Variations in thermal conditions and rolling speed (interpass time) was used to affect the recrystallisation process during rolling (full and particular recrystallisation).
In the rolling trial two different sets for the rolling speed were used. The related sets of interpass time are given in Table A5.1. In [A5.1] the variations of the initial thermal conditions are reported.
During the rod rolling trials parameter were varied to study the effect of:
Reheating temperature,
Reheating time (at const. Temperature),
Page 316
Rolling speed,
Temperature at the entrance of F1.
Strain variations were done due to the dependence of spreading on the rolling speed and temperature (const. roll gap geometry see [A5.1]).
From a plot of the rolling forces v time it is known that a stationary state was reached. Dependent on rolling speed and rolling temperature three principal trends of rolling forces were found (trend13 in Figs. A5.7 and A5.8). But all trials plotted in Fig. A5.7 results in similar final grain sizes (Table A5.2).
In general it is to be seen in the rolling force plots that the rolling speed and average temperature affects the recrystallisation process in a complex manner. But microstructural investigations of the material after the last pass (F4) had shown no significant differences in the final average austenitic grain size. Thus more detailed investigations of the microstructure after F1 and F2 are necessary to quantify the influence of the parameter given above.
The qualitative effect of reheating temperature and time can be derived from the development of the rolling forces during the whole rolling process:
The trend curves in Fig. A5.9 represent the principal curves 1 (rolling speed 5 m/s) and 2 (rolling speed 30 m/s) in Figs. A5.7 and A5.8.
A higher reheating temperature results in lower rolling forces for constant rolling speed.
All trials plotted in Fig. A5.9 started at 850°C in the first pass F1. For different reheating temperatures similar cooling times (26 s … 28 s) were necessary to reach 850°C. Due to different reheating temperatures grain growth reaches different stages, for specimen reheated up to 1200°C a more homogeneous microstructure but also more scale occur. For the freecutting steel used the scale forces lubrication during rolling. So finally the reason for a lower rolling force can’t be attributed only to grain growth. Therefore a detailed analysis of the microstructure obtained after the first two passes are necessary. Computation of a single pass and comparison with experimental results will also give some information about the significance of friction and recrystallisation.
During the variation of the rolling speed the ration of the speed between the stands was held constant. The effect of speed variations is to be seen in Figs. A5.9 and A5.11.
At constant thermal initial conditions (Fig. A5.11) different trends in the rolling forces were obtained. Therefore only particular recrystallisation can occur. The trends in temperature (at F2F4) for the trials given in Fig. A5.11 are similar but different in its absolute value. A possible reason is the shortening of the interpass time with increasing the rolling speed (Table A5.1).
Page 317
At constant rolling speed the temperature at the entrance will also affect the trend of the rolling forces (Fig. A5.10). The rolling forces not only increase with higher temperatures. The trends differs also in the inclination inbetween the passes. Similar results available at lower speed (see Fig. A5.7). This behaviour can be attributed to the material behaviour in the transformation range.
While differences in rolling force are visible, the final grain size (behind F4) will be only slightly affected by all parameter changes. Smaller grain sizes can be measured for higher reheating temperatures or higher reheating times and higher rolling speed at lower temperatures in the entrance of F1.
For the final evaluation mean values of the rolling forces were used. Due to a short maximum length of the specimen compared to the contact length small regions of a stationary state were observed in the first pass.
In Fig. A5.19 rolling forces v rolling temperatures are plotted. Only an increase of the rolling forces with decreasing temperatures is to be seen. Other information couldn't derived from this kind of plots. For comparative studies dimensionless plots will be derived to exclude the influence of deviations in roll geometry and temperature. After [A5.6] and [A5.5] the ratio of the mean deformation resistance kWm and mean yield stress kFm is plotted in dependence on the ratio of the contact length Ld and the mean thickness of the roll gap hm (Fig. A5.24).
[VER FORMULA EN PDF ADJUNTO]
To derive the basic ideas the analogy of upsetting test and rolling can be used: The pressure distribution and thus the force depend on the ratio of contact length and thickness of the specimen, the friction stress and the mean yield stress.
Under an aspect of similarity the rolling force F is a function of the rolling geometry, rolling speed, frictional forces and yield stress only. The effect of temperature and strain are included in these parameters. Thus we will set:
[VER FORMULA EN PDF ADJUNTO]
Using the Pi – theorem ([A5.7]) in dimensionless form we are able to reduce the number of independent entities by 3. To rewrite this relation for constant rolling speed only 4 dimensionless entities are necessary:
[VER FORMULA EN PDF ADJUNTO]
Page 318
[VER FORMULA EN PDF ADJUNTO]
Figure A5.25 shows a typical plot for the dependence of the normal pressure und thus the change of the rolling force on the aspect ratio Ld /hm.
For comparative studies the mean forming resistance kWm is given from experimental data by
[VER FORMULA EN PDF ADJUNTO] with the width of rolling specimen B. The mean value of the yield stress kFm will be computed for each pass by
[VER FORMULA EN PDF ADJUNTO]
In Fig. A5.24 the relation of the ratio kWm/kFm to the aspect ratio Ld /hm is given.
Two main parts of interest are to be seen. An interpretation is easy derived from elementary theory of plasticity:

[VER FORMULA EN PDF ADJUNTO] In this range the forming resistance is mainly affected by geometric conditions and the yield stress. Deviations in the stressstrain behaviour but in geometrical conditions also will result in significant deviations in kWm/kFm.

[VER FORMULA EN PDF ADJUNTO] : In this range of Ld/hm the ratio kWm/kFm depends roughly linear on Ld/hm. From elementary theory of plasticity it is known friction forces are dominant in this parameter range. The trials were performed in this region. Therefore during the trials attention was taken on a complete and homogeneous descaling process after reheating.
In general for plate rolling without tensional stresses at the entrance or exit of the roll gap the ratio kWm/kFm will depend on the following parameter [A5.5]:
[VER FORMULA EN PDF ADJUNTO]
The length scales are given above, whereas the frictional coefficient µ will increase in the hot forming region with increasing normal pressure, decreasing rolling temperature and strain rate. A stress strain behaviour and thus the mean yield stress of the pass will depend on:
[VER FORMULA EN PDF ADJUNTO]
On the basís of these discussions the results of the rolling trials will be analysed below:
Page 319
In Fig. A5.26 the data of the 1 and 2 pass are plotted in dimensionless form. Additional parts of the upper and lower curve from Fig. A5.24 are inserted. The data of the 1 passes of both rolling series (30 mm and 7 mm initial thickness) are located near the lower curve.
These results will prove again the basic assumptions of the similarity model.
Observed deviations in the 1 pass will result in deviations of the computation of the mean width, different cooling conditions and thus a change of the frictional stresses and mean yield stress. Due to a temperature gradient the specimen will be at a higher temperature in the core compared to surface. Thus the computed mean yield stress will be higher. The same effect will result from the approximation of the stressstrain curves. At a strain of 0.6 and more the model will give higher values than the experimental data (+5…20 MPa) at the strain rate and temperatures used. With increasing temperature the frictional stresses will also decrease. These effects will result in a lower ratio kWm/kFm.
In general the results in the 2 pass are located at the upper boundary or higher in Fig. A5.28. Data obtained in the 1 trials with initial thickness of 30 mm will give higher values for kWm/kFm than data from the 2 series. Furthermore differences in the ratio of kWm1/kFm1 and kWm2/kFm2 of the 1 and 2 pass were predicted (Fig. A5.27).
For a similar aspect ratio [VER FORMULA EN PDF ADJUNTO] in the second pass the dimensionless ratio kWm2/kFm2 of the trials using 30 mm initial thickness shows higher values than for trials with 7 mm initial thickness. Both series differ in strain history and aspect ratio B/h0 (Table A5.3 and A5.4, Figs. A5.26 and A5.27).
The mean yield stress was calculated using the Freiberg model equation for the continuous stress –strain curves were the softening behaviour due to interpass time was neglected. In the future work discontinuous upsetting test will follow.
To identify the role of strain history and recrystallisation we have to discuss the effect of geometry and friction first.
From (7) the ratio kWm/kFm will increase for a constant aspect ratio Ld/hm for decreasing temperature, aspect ratio B/h0 and increasing ratio of real to computed stress – strain – behaviour:
A decrease in rolling temperature will result in increasing frictional stresses. This will result an increasing ratio kWm/kFm. Different interpass times result in different frictional conditions due to the growth of oxide scale of different thickness.
A decreasing aspect ratio B/h0 for the different series and thus in the stress state (2D or 3D) have to result in a reduced ratio kWm/kFm. From experimental data at constant [VER FORMULA EN PFD ADJUNTO] we obtain a contrary behaviour (Fig. A5. 26). Thus comparing both rolling series the aspect ratio B / h0 is not the main parameter. Both series differ in the strain history (Fig. A5.27). In the 1 series with initial thickness 30 mm significant increase in the ration kWm/kFm is obtained. This is on the one hand due to a rise in [VER FORMULA EN PDF ADJUNTO] but also due to a significant change in the stress state [VER FORMULA EN PDF ADJUNTO]. In the 2 series with initial thickness 7 mm in thePage 3201 pass a two dimensional stress state is reached [VER FORMULA EN PDF ADJUNTO].Thus a further increase in B/h0 will not significant affect the stress state. No increase of the width was measured for [VER FORMULA EN PDF ADJUNTO] but also for one trial in the 1 series with [VER FORMULA EN PDF ADJUNTO] and a strain of 0.35 in the second pass similar to the second series. This single trial gives a higher kWm/kFm in the second pass too. This will be an indication of the influence of friction and stressstrainbehaviour.
Changes in the real stress  strain – behaviour will affect the mean yield stress and thus the value of kWm/kFm because these behaviour was assumed unchanged for the calculation of kFm. For these changes a few reasons are possible:
(i) Deviation between measured and real surface temperature and between surface and core as well.
(ii) Softening processes:
For a 100% removal of work hardening at an interpass time of more than 5 s like in the rolling trials will be enough at 1000°C. Below 950°C an inhibition of static softening due to forming induced precipitations will occur.
Grain refining due to recrystallisation in combination with static softening in the 1 pass will result in a steeper increase of the stress strain curve in the 2 pass compared to curves for a larger grain size. This results in a higher real kwm in the 2 pass. Whereas the 1 series were performed in the region of dynamic recrystallisation ([VER SIMBOLO EN PDF ADJUNTO] in the 1. pass, Fig. A5.27) the trials of the 2 series were performed without full DRX. Therefore higher values of kwm/kFm in the 1 series are possible.
Grain growth e.g. at 1000°C after a full RX is a process working in the opposite direction of grain refining and lower the ratio of kwm/kFm. Higher strain rates und strain will force grain growth at constant interpass time. The stress – strain curve of such trials will be close to a continuous curve. Particular this effect will be possible for the first series with initial thickness of 30 mm at 1000°C (Fig. A5.27).
An inhomogeneous distribution of real strain and strain rate and strain history may also affect the ratio kwm/kFm.
Microstructural investigation:
In the plate rolling trials several specimen of similar geometry and similar initial microstructure where reheated at the same temperature and same time. Then the specimen where rolled at the same rolling geometry in the first and second pass. The difference between both groups of trials was the rolling direction in the second pass. Therefore different microstructures where predicted.
From microstructural investigations of the rolled specimen we obtain a smaller grain size after reversal rolling than for forward rolling (Figs. A5.28 and A5.29). This behaviour isPage 321independent of the rolling temperature. An inhomogeneous grain size distribution was obtained. The general differences could be due to local differences in the strain pass history.
In several FE computations using the FE code MARC including subroutines for the stress strain behaviour the rolling trials described in A5.2 were analysed (Figs. A5.34 and A5.35). Due to differences in the shear deformation during the second pass deviations between forward and reversal rolling were obtained. Furthermore the flow lines during the passes were computed. These results were used for comparison with results obtained in Visio plasticity measurements.
The method of visioplasticity was optimised in several rolling trials for online measurements in hot rolling. The aim was to predict the real flow behaviour in dependence of the rolling direction and to exclude frictional behaviour.
From computational studies the origin of the inhomogeneity and the differences in the RX behaviour was identified.
The experimental results were used for comparison with FE  simulations. In Fig. A5.36 – Fig. A5.38 results obtained in rolling trials and 2D – FE simulations are shown. Isotropic hardening and von Mises material where used in FE – computations. The deformation of the flow lines and the wire grid show similar results. Small deviations are to be seen near the exit of the rolling gap in the middle of the rolling sample as well as in an area of ¼ of the sample thickness below the surface.
The origin of the deviations is the out  of  plane displacement of the specimen surface (Fig. A5.38(b)). Other reasons are frictional effects and the influence of thermal boundary conditions. Further simulations using 3D FE –Models will help to overcome these deviations.
A5.1 Technical report No.1 of the ECSC Project 7210.PR/291 (D3.05/01); “Constitutive modelling complex loading in metal forming processes”, March 2002.
A5.2 Technical report No. 2 of the ECSC Project 7210.PR/291 (D3.05/01);“Constitutive modelling complex loading in metal forming processes”, August 2002.
A5.3 Technical report No. 3 of the ECSC Project 7210.PR/291 (D3.05/01); “Constitutive modelling complex loading in metal forming processes” , March 2003.
Page 322
A5.4 Gutierrez, I, Lopez, B and RodriguezIbabe, J M: “Grain growth phenomena duringsteel processing”, in: Proc. 1. Joint International Conf. Recristallisation and Grain Growth, Gottstein, G; Molodov, D.A., 2001, pp145 – 154.
A5.5 Hensel, A (ed.): „Technologie der Metallformung – Eisen und Nichteisenwerkstoffe“, VEB Deutscher Verlag für die Grundstoffindustrie Leipzig, 1990.
A5.6 Hensel, A and Spittel, T: „Kraft und Arbeitsbedarf bildsamer Formgebungsverfahren“, VEB Deutscher Verlag für die Grundstoffindustrie Leipzig, 1978.
A5.7 Spurk, J H: „Dimensionsanalyse in der Strömungsmechanik“, Springer Verlag, 1992.
Page 323
Table A5.1: Interpass time in continuous rod rolling
vF4  F1F2  F2F3  F3F4  F4Cooling 
5.39m/s  0.6s  1.1s  0.26s  1.17s 
32.36m/s  0.1s  0.2s  0.04s  0.19s 
Table A5.2: Average austenitic grain size of rod rolling trials
[VER TABLA EN PDF ADJUNTO]
Page 324
Table A5.3: Rolling conditions for plate rolling (1 Series) Rolling speed 1 m/s; diameter of roll 360 mm
[VER TABLA EN PDF ADJUNTO]
(Continued)
Page 325
(Continued)
[VER TABLA EN PDF ADJUNTO]
Page 326
Table A5.4: Rolling conditions for plate rolling (2 Series) Rolling speed 1 m/s; diameter of roll 380 mm
[VER TABLA EN PDF ADJUNTO]
(Continued)
Page 327
(Continued)
[VER TABLA EN PDF ADJUNTO]
Page 328
Table A5.5: Calculation constants for stress – strain curves
[VER TABLA EN PDF ADJUNTO]
Page 329
Table A5.6: Calculation constants for stress – strain curves
[VER TABLA EN PDF ADJUNTO]
Page 330
Fig. A5.1: Structure of work program
[VER GRAFICO EN PDF ADJUNTO]
Page 331
Fig. A5.2: Scheme of benchmark trials and steel grades used
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.3: Continuous high speed rolling mill (water bath fro trials F1F2; F1 only; layer for high speed rolling only)
[VER GRAFICO EN PDF ADJUNTO]
Page 332
Fig. A5.4: Roll pass design
[VER GRAFICO EN PDF ADJUNTO]
Page 333
Fig. A5.5: Stands F1 (right) and F2 (left) for rolling F1F2 at 5 m/s
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.6: Configuration of plate rolling trials (a: single pass; b: reversal rolling; c: forward rolling 2 passes)
[VER GRAFICO EN PDF ADJUNTO]
Page 334
Fig. A5.7: Rolling forces during the continuous rod rolling (selection)
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.8: Rolling temperatures during the continuous rod rolling (selection)
[VER GRAFICO EN PDF ADJUNTO]
Page 335
Fig. A5.9: Effect of reheating temperature and rolling speed on rolling forces during the continuous rod rolling – const.
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.10: Effect of temperature at the entrance of F1 on rolling forces during the continuous rod rolling at constant rolling speed (30 m/s)
[VER GRAFICO EN PDF ADJUNTO]
Page 336
Fig. A5.11: Effect of rolling speed on rolling forces (const. Reheating temperature, reheating time and temperature at entrance F1) during the continuous rod rolling
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.13: Austenitic grain size evolution during heating for different steel grades [A5.4]
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.14: Abnormal grain growth in a CMnNb grade; 20 min 1100°C [A5.4]
[VER GRAFICO EN PDF ADJUNTO]
Page 337
Fig. A5.12: Austenitic grain size at the end of pass F4 for different rolling conditions (see [A5.1]; Fig. A5.3 and average grain size Table A5.2)
[VER GRAFICO EN PDF ADJUNTO]
Page 338
[VER GRAFICO EN PDF ADJUNTO]
Page 339
Fig. A5.15: Compted t8.5 time for different thickness of plate
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.16: Phase transformation behaviour of CMnNb steel
[VER GRAFICO EN PDF ADJUNTO]
Page 340
Fig. A5.17: Cooling test of plate rolling specimen thickness 15.6 mm
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.18: Cooling test of plate rolling specimen thickness 7.5 mm
[VER GRAFICO EN PDF ADJUNTO]
Page 341
Fig. A5.19: Final geometry and microstructure of specimen in the 1 series (initial thickness 30 mm) after the 2 pass
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.20: Final geometry and microstructure of specimen in the 2 series (initial thickness 7 mm) after the 2 pass
[VER GRAFICO EN PDF ADJUNTO]
Page 342
Fig. A5.21: Rolling forces and phase transformation behaviour (lower curve: yield stress 10/s; M= 0.2; A3: 750°C … 800°C)
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.22: Yield stress of CMnNb steel at fixed strain (0.2)
[VER GRAFICO EN PDF ADJUNTO]
Page 343
Fig. A5.23: Yield stress of CMnNb steel at fixed strain (0.5)
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.24: Dimensionless forming resistance kWm/kFm v aspect ratio Ld/hm [A5.5]
[VER GRAFICO EN PDF ADJUNTO]
Page 344
Fig. A5.25: Geometry for plate rolling (left) and its effect on normal pressure distribution (right) [A5.7]
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.26: Dimensionless forming resistance kWm/kFm v aspect ratio Ld/hm for 1 and 2 series of plate rolling
[VER GRAFICO EN PDF ADJUNTO]
Page 345
Fig. A5.27: Ratio of dimensionless forming resistance kWm1/kFm1 of the 1 pass v kWm1/kFm1 of the 2 pass (1 and 2 series)
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.28: Ratio of dimensionless forming resistance kWm1/kFm1 of the 1 pass v kWm1/kFm1 of the 2 pass (2 series (7 mm) only)
[VER GRAFICO EN PDF ADJUNTO]
Page 346
Fig. A5.29: Microstructural evolution during 2 passes in dependence on the rolling direction
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.30: Microstructural evolution during 2 passes in dependence on the rolling direction
[VER GRAFICO EN PDF ADJUNTO]
Page 347
Fig. A5.31: Fractional softening for monotonic compression test: 1000°C From CEIT
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.32: Fractional softening for monotonic test: 0.1/s; 1150°C after CEIT
[VER GRAFICO EN PDF ADJUNTO]
Page 348
Fig. A5.33: Fractional softening of monotonic test: 1/s; 1100°C after CEIT
[VER GRAFICO EN PDF ADJUNTO]
Fig. A5.36: FE computations: total equiv. plastic strain and flow lines for the rolling trial W3 – centre line is symmetry line
[VER GRAFICO EN PDF ADJUNTO]
Page 349
Fig. A5.34: FE computations: total equiv. plastic strain and flow lines for the 1st pass of forward rolling (a) and reverse rolling (b)
[VER GRAFICO EN PDF ADJUNTO]
Page 350
Fig. A5.35: FE computations: total equiv. plastic strain and flow lines for the 2nd pass of forward rolling (a) and reverse rolling (b)
[VER GRAFICO EN PDF ADJUNTO]
Page 351
Fig. A5.37: Rolling trial W3 with marked surface for visioplasticity investigations
[VER GRAFICO EN PDF ADJUNTO]
Page 352
Fig. A5.38: Automatically obtained displacement increments for two positions during the rolling trial W3 with marked surface