Conclusions and recommendations

AuthorG.J. Watts- D.C.J. Farrugia- B. Cheong- Z. Husain- M. Zhou I. Gutierrez- D. J. Badiola- J. H. Bianchi- P. Vescovo O. Wiklund- M. Karlberg- M. Schmidtchen- R- Kawalla L. P. Karjalainen- M. C. Somani
Pages33-39

Page 33

The objectives of the project, in conformity with the Technical Annexe, have been achieved through the sharing of expertise, helpful co-operation of all the partners and synergies resulting therefrom. Detailed conclusions regarding partners' activities are given in Appendices 1 to 6. General conclusions regarding the tasks are given below.

4. 1 Modelling

A review of constitutive models has shown that, generally, those previously employed are intrinsically unsuitable for providing a comprehensive understanding of the influence of complex loading paths since they do not include the coupling between local quantitiesPage 34describing or closely affected by the structural behaviour and the thermomechanical variables. They may however be useful in providing a partial understanding of the effects of path e.g. on shape and in assisting with designing schedules etc.

Modelling the influence of deformation history on macroscopic quantities such as shape can be carried out, without including the structural quantities, either by skillfully incorporating information, gleaned from observation, into the input data of the model, updating the latter as necessary or using more complex forms of constitutive equation (e.g. of integral of functional type). The first approach has been adopted by TU Freiberg, by means of online visioplasticity in conjunction with a correlation based image processing technique and has obtained close predictions of the effect of reversal in the plate rolling trials.

Modelling of the structural coupling can be achieved either by a 'bottom-up' approach in which the structural behaviour throughout the stock is described using a microstructural model, to which mesoscopic variables such as stress are related or, by a 'top-down' approach in which a multi-level model is progressively developed from a standard mesoscopic model by the successive augmentation of the set of variables to contain quantities closely related to the structural evolution. The first approach, in principle, provides a comprehensive description of the processes in the stock but requires extensive and expensive acquisition and processing of data. The second approach has a well defined but limited scope, can predict associated quantities relatively easily with a moderate amount of data and can be extended to other levels.

The first approach has been adopted by MEFOS in the development of a sophisticated fundamental model based on dislocation dynamics, vacancy generation and recrystallisation including nucleation based on abnormal grain growth (developed in conjunction with SIMR and MIKRAB) from which the flow stress is derived by a modified Hirsch equation and incorporated into an elastoplastic FE model. Many data for yield stress over a range of strains, strain-rates and temperatures, etc. have been collected for CMnNb from Freiberg and CCrV steel from SIMR. Problems were encountered in choosing appropriate coefficients of the constitutive model because of difficulties in fitting a non-analytical (i.e. integrated from a system of differential equations) function to experimental curves, possibly reflecting limitations in state of the art optimisation methods. Predictions of stress, strain etc. and the microstructural quantities have been obtained for a compression test (to validate the model) and also for the benchmark application of plate rolling at Freiberg. Reasonable predictions of the thermomechanical variables have been obtained but verification of structural quantities such as dislocation density was not possible.

A similar approach, but involving fewer variables, was followed by Corus in adopting the dislocation based model developed in conjunction with Birmingham University. Unlike previous models, it efficiently integrates the constitutive equations and predicts stress, strain, grain size evolution, fraction recrystallised and dislocation density. It has been applied to the H-V rolling trials of Type 316 Stainless steel using data obtained from delayed double hit tests on the Gleeble. It predicts very little recrystallisation on entry to pass two as indicated by the experimental rolling trials.

The second approach has been followed by CSM in the development of a model describing hardening, dynamic recovery, dynamic recrystallisation and static softening. A major aim of the model is to investigate differences between shear reversal and normal reversal in hot rolling. It has been applied to rolling of both a fast recrystallising medium carbon steel and a non-recrystallising duplex stainless steel using torsion and compression data from CEIT and the University of Oulu to determine the coefficients in the constitutive equations.

Page 35

The model has been validated using the above facilities even for conditions of changing strain rate.

The model has also been validated on its local plate mill and on the ORI –MARTIN bar mill and gives rolling load and power predictions in agreement with measured values.

In spite of differences in shear strain accumulation, minimal differences in roll force between roughing and tandem configurations are predicted for the nonrecrystallising steel.

For bar rolling of medium C steel, the model predicts DRX at the core unlike simplistic models, which predict SRX everywhere.

It has not been possible to verify the structural predictions metallographically.

Softening does occur in practice but not to the extent predicted by the model assuming full softening of the yield surface.

Useful non-numerical models describing the effect of strain have also been developed in the project.

An empirical set of equations describing the entire flow stress curve for strain reversal has been established by CEIT.

An analytical model of the flow stress curve following a change in strain-rate, based on solving a dislocation model of MEFOS with simplifying assumptions has been obtained by the University of Oulu and graphs obtained are in good agreement with its experimental results.

An analytical model for the fraction recrystallised, including the effects of strain reversal has been derived, by CEIT, from the underlying theory based on dislocation density and grain boundary misorientation and gives approximately good predictions of softening except at the end of the plateau.

In application and further development of the models it is recommended that

More comparison is made between predictions and measurements of microstructure and development of more accurate techniques for the latter is actively pursued.

Techniques are developed for the efficient and accurate numerical integration of the constitutive equations.

State of the art optimisation techniques are explored in order to determine coefficients which achieve a closer fit of the integrated solution of the constitutive equations to the experimental flow curve.

Page 36

4. 2 Experimental investigation and material characterisation

There is evidence from experimental trials that the complex loading paths considered, for certain values of the process parameters, had a significant influence on the flow stress behaviour and the microstructural evolution:

In tension tests, an abrupt decrease in strain rate, in the dynamic recovery regime, gives rise to;

- A lag in flow stress behind the behaviour of a mechanical equation of state.

- An even greater lag in the static recrystallisation rate, especially after a strain resulting in recrystallisation (the values for the above close to, or higher than, those for the constant initial strain rate).

- For ferrite, larger cells than those if the initial strain rate is maintained.

In tension/compression and torsion, a strain reversal can give rise to;

- A noticeable Bauschinger effect.

- A transient characterised by a plateau.

- A change in the SRX rate depending on the magnitude of the strains and for sufficiently large interpass times.

- For ferrite, grain coarsening, larger cells and more low angle boundaries.

In tension/compression and torsion, a strain reversal coupled with an abrupt drop in strain rate gives rise to;

- A more pronounced Bauschinger effect.

- A longer plateau.

In torsion, a strain reversal coupled with an abrupt increase in strain rate gives rise to the opposite of the above effects for a decrease (Note: no relevant data available in this project for tension/compression).

In tension/compression, re influence of temperature on austenitic steels;

- Thermal history effects are insignificant (viz. a slight increase in flow stress for a rapid increase in temperature from 900°C to 1100°C).

- Actual interpass temperature does affect recrystallisation.

Multidirectional compression tests indicate that;

- Multiple hits at constant temperature give slower early recrystallisation than a single hit of the same total magnitude and may inhibit REX entirely if, individually, are sufficiently small.

Page 37

- An interpass turn of 90° hastens early recrystallisation when one of the reductions is large but significantly retards it when both hits are moderately large.

Even though results from experimental tests need to be qualified according to the parameters employed, significant effects of loading path have frequently been observed. In contrast, effects from rolling trials have generally been found to be secondary and in some cases, the evidence has been inconclusive or even conflicting.

In forward –reverse rolling trials using CMnNb steel plates;

- Noticeable differences between the deformation of stock rolled in the same direction and that for the stock reversed have been observed using visioplasticity techniques.

- Positive and negative changes in load in 2nd pass for the two schedules have been observed depending on the temperature.

- Differences in development of austenitic grain structure have generally been difficult to detect because of the presence of bainite etc, but, where martensite was produced, the evidence was that reverse rolling tended to reduce the grain size.

In H-V rolling trials using Type 316 stainless steel bars;

- The fraction recrystallised locally i.e. near the surface is smaller in stock that has undergone an interpass turn but the load following the interpass is smaller suggesting that recrystallisation is not widespread.

- In trials investigating the effect of various interpass heating programmes with or without an interpass turn, optical metallography could produce no reliable evidence of significant or absent recrystallisation for both the interpass heating designed to promote SRX and that designed to inhibit it. Electron microscopy using EBSD produced IQ maps of poor quality and was therefore inconclusive. Pole figure maps, however, showed slight evidence of a little recrystallisation in both cases. OM also revealed wide variation in grain size and show that maximum grain sizes are greater in specimens that have been turned; the large differences however are more likely indicative of material inhomogeneity than retardation of grain refinement.

In high speed rod rolling;

- Slightly smaller grain sizes have been observed at the higher speeds and higher reheating temperatures.

- More significantly, loads are higher at higher temperatures-probably because of phase transformation.

Inconclusive results from rolling trials could have arisen because;

Page 38

Effects were masked by other phenomena in particular related to heating and cooling.

Difficulties in measuring the quantities required in relation to;

- Measurement technology.

- Material employed.

Statistical variation.

The deformation paths considered in a given trial are not sufficiently different from one another:

- The strain rate changes after the roll bite are not as large as those considered in the tests.

- Despite the apparent difference, the strain angle between the deformation paths is not large for the process considered – as remarked in the introduction.

To address these issues in particular, differences between results from experimental tests and rolling trials, the following recommendations are therefore made in implementation of suggestions from the project and in future research.

Reheating and cooling should be carefully controlled and monitored by integration and automation of heating, rolling and quenching facilities comparable to the set up in Gleeble machines. Instrumented rolling trials should be carried out and specially constructed refractory units should be used where approximately isothermal conditions are required.

Difficulties in measuring quantities should be reduced by;

- Use of more sophisticated technology such as EBSD etc using field emission rather than tungsten filament equipment, SEM etc.

- Use of materials e.g. Fe30% Ni, with analogous mechanical properties to those of practical interest but more amenable to observation but without obfuscating phases from transformation and significant initial heterogeneity.

Statistical variation should be reduced by; (a) the use of more samples taken from more positions of the rolled stock, for metallographic examination; (b) the taking of a more statistically meaningful sample by repetition of trials and the consideration of other combinations of parameters selected using DOE techniques.

More complex loading paths occur in practice globally and locally during rolling and should be considered, in particular

- Greater strain rate variation e.g. effects of roll acceleration/deceleration and skidding etc.

Page 39

- Greater changes in strain path both globally e.g. in edge rolling, 'cross rolling' operations for plates and locally in the rolling of beams near the web-flange interface and in sections such as piling near the lock . The latter can be studied using FE and then reproducing the deformation paths predicted, in samples using an ASP machine to examine the structural evolution.

It is believed that the choice of loading path is important e.g. as a means of refining the grain size by recrystallisation and thereby increasing the strength. This could be important for plates and beams as an alternative to heat treatment or to increasing the carbon content which is detrimental to weldability. The loading path may also be significant when used in conjunction with novel or composite rolling shapes such as may be relevant in downstream applications, and with other material qualities more susceptible to changes in process history.

The project suggests means for studying experimentally and theoretically the effects of loading path, a study which should continue, even if down a different avenue.

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT