mechanical considerations - Lehrstuhl für Werkstoffkunde und
Transkript
mechanical considerations - Lehrstuhl für Werkstoffkunde und
INTERNATIONAL DOCTORAL SEMINAR, May 16 – 19, 2010, Smolenice castle MICROSTRUCTURE ANALYSIS OF A BANDED WORK ROLL WITH COMPARISON BETWEEN DAMAGE APPEARANCE AND THERMOMECHANICAL CONSIDERATIONS 1 1,2 2 3 M. DÜNCKELMEYER , C. KREMPASZKY , E. WERNER , G. HEIN , 3 K. SCHÖRKHUBER 1 Christian Doppler Laboratory of Material Mechanics of High Performance Alloys, Technische Universität München; Boltzmannstraße 15, 85747 Garching, Germany 2 Chair of Materials Science and Mechanics of Materials, Technische Universität München; Boltzmannstraße 15, 85747 Garching, Germany 3 voestalpine Stahl GmbH, voestalpine-Straße 3, 4020 Linz, Austria ABSTRACT Banding is a mechanism which degrades the surface of work rolls for hot rolling due to thermo-mechanical overloads. In order to investigate the cause of banding, material taken from a banded work roll has been analyzed by light optical microscopy (LOM) and scanning electron microscopy (SEM) with focus on the transition between banded and un-banded regions. The mechanisms of crack formation in carbides and in the matrix material are investigated with respect to the direction of loading. The observed banding depth is compared with an analytically calculated plastic zone size associated with thermally induced stresses. Key words: Hot rolling, Banding, Roll damage, Thermo-mechanically induced damage INTRODUCTION The performance of work rolls for hot rolling is governed by several degradation phenomena, such as mechanical fatigue, wear, thermal fatigue, and oxidation. Therefore much effort has been undertaken to describe and classify thermo-mechanically induced roll damage like banding [1]. DEBARBADILLO et al. [2] define banding as the surface deterioration of early finishing stand work rolls brought about by the tearing out of very small pieces of roll material, see Figure 1. WALMAG et al. [3] report that in a common rolling mill 0.6 % of coils are rejected due to rolled-in work roll scale which is caused by banding. An optimisation of the hot rolling process by means of changed process parameters in order to reduce banding is discussed by ARNCKEN et al. [4]. REIP [5] shows, that the main reason for banding of work rolls is the high temperature gradient present in the rolling gap. ERICKSON et al. [6] conclude that the thermo-mechanical fatigue of the roll surface due to continuous temperature cycling and resulting mechanical stresses are mostly the cause of wear associated with banding. Therefore the temperature distribution in the work roll und thermally induced stresses during hot rolling have been investigated by several authors [7-11] with the help of standard finite element (FE) solvers. DÜNCKELMEYER et al. [12] propose an analytical approach to investigate the instationary 126 INTERNATIONAL DOCTORAL SEMINAR, May 16 – 19, 2010, Smolenice castle temperature distribution and the resulting stresses in circumferential direction of the work roll. Due to its advantages over numerical parameter studies, the approach will be used in this work in order to correlate experimentally observed damage with theoretically estimated plastic deformation. Damage is documented with detailed metallographic analyses of the work roll surface and profile sections of a banded roll segment. Crack formation will be discussed with respect to damage mechanisms on the microscale. INVESTIGATED MATERIAL In order to tolerate high thermal and mechanical loads work rolls of a hot rolling mill have to fulfill high demands regarding their properties. For this reason mainly High Chromium Steel (HiCr) as well as High Speed Steel (HSS) rolls are used in the first stands of a rolling mill. In this contribution a segment of a HiCr roll with a banded surface is investigated. DEBARBADILLO et al. [2] showed that the sequence of events which characterize banding failure is the same regardless of the roll material, even though HSS work rolls seem to withstand banding better than HiCr work rolls. The chemical composition and the phases present in the microstructure of the investigated HiCr work roll material are shown in Tables 1 and 2. Primary carbides occur clustered, whereas secondary carbides are finely distributed in the microstructure. Table 1 Chemical Composition C Si Mn Cr Ni Mo V W 2.82 % 0.63 % 0.95 % 14.84 % 1.32 % 1.13 % 1.15 % 0.04 % Fig. 1 Surface damage of an industrially used work roll Table 2 Phase content (volume fraction) Martensitic Primary Secondary matrix carbides carbides 65 % 22,5 % 12,5 % Fig. 2 Investigated HiCr roll segment with banded surface Fig. 3 Views directions MICROSTRUCTURAL OBSERVATIONS Figure 2 shows the cutting pattern applied to the investigated HiCr roll segment. The specimens are wire-guided electrical discharge machined and then embedded in epoxy-resin. After grinding and polishing, specimens for light microscopy are etched with Nital. The transition zone between banded and un-banded areas is investigated in longitudinal, lateral and top views, see Figure 3. Figure 4 shows an overview of the transition between banded and un-banded areas in longitudinal view. Carbides (mainly Cr7C3 [13, 14]) appear dark in the brighter martensitic matrix. Since the work rolls are fabricated by centrifugal casting, carbides are mainly elongated in radial direction [15]. Near the banded area – which can be identified macroscopically as non-shining dark surface, see Figure 1 – the un-banded 127 INTERNATIONAL DOCTORAL SEMINAR, May 16 – 19, 2010, Smolenice castle area shows surface near damage in form of a crack network. Cracks, perpendicular to the surface – so called fire cracks - with lengths up to 120 µm can be detected. The thickness of the removed layer increases until a maximum depth of approximately 150 µm is reached. Fig. 4 Damage of a work roll in an overview of the transition between banded and un-banded areas in longitudinal view, SEM Figure 5 shows carbides fractured parallel to the roll surface in the banded region in lateral view. A strongly damaged region can be seen in the lower left corner of the image. It can be imagined that this region is on the verge of its separation from the roll. Figure 6 shows the transition zone between banded and un-banded regions in top view. Wear marks due to the contact between the roll and the strip in the rolling gap can be identified in the banded area. Figure 7 shows the microstructure in longitudinal view. The coalescence of cracks - parallel and perpendicular to the surface – and the adhesion of the deteriorated layer to the strip in the rolling gap cause its separation from the roll. A progressed crack network is detected. Figure 8 shows the transition of banded and un-banded region in longitudinal view. All microscopic observations show that regions with a high density of carbides also exhibit a pronounced crack network. These regions are predestined for the initial delamination of a thin surface layer. A fire crack with a depth of 120 µm can be seen in the center of the image. Figure 9 shows the relation between the banded and un-banded roll surface in lateral view. From this image a maximum banding depth of approximately 150 µm can be measured. Fig. 5 Carbides fractured parallel to the roll surface (dark) and martensitic matrix (light), lateral view, SEM. Fig. 6 Transition zone between un-banded and banded areas in top view, SEM. 128 INTERNATIONAL DOCTORAL SEMINAR, May 16 – 19, 2010, Smolenice castle carbide matrix Fig. 7 Carbides fractured parallel to the roll surface (light) and martensitic matrix (dark), longitudinal view. LOM, Nital etching. Fig. 9: Fig. 8 Transition zone between un-banded and banded areas in longitudinal view, SEM. Banded area in lateral view, LOM THERMO-MECHANICAL CONSIDERATIONS The high temperature gradient due to the contact between roll and strip in the rolling gap causes high residual stresses in surface near layers of the roll and is therefore responsible for the deterioration of the roll surface [5]. Hence it is useful to know temperatures and residual stresses in the work roll. In the following an established analytical macroscopic model [12] is employed in order to correlate an estimated theoretical penetration depth defined as depth to which thermally induced yielding of the work roll material occurs - with the experimentally observed thickness of removed layers. The analytical model is structured as follows: Calculation of a) maximum temperature of the roll, b) temperature distribution in the roll, and c) residual stresses. Stresses due to the rolling pressure in the rolling gap are neglected since they are 15 times smaller than the thermally induced stresses [5]. The temperature of the roll at the exit of the rolling gap can be described as [16] Troll { ( )} K1κ 1−0.5T0 h22κ 2t = 1 − e erfc h2 κ 2 t , K1κ 1−0.5 + K 2κ 2−0.5 (1) where indices 1 and 2 denote the strip and the roll, respectively, K and K are the thermal conductivities, κ and κ are the thermal diffusivities, erfc is the complementary error function, t is the contact time, T0 is the initial temperature of the strip and the quantity h2 is defined as, 1 1 2 129 2 INTERNATIONAL DOCTORAL SEMINAR, May 16 – 19, 2010, Smolenice castle H ( K1κ 1−0.5 + K 2κ 2−0.5 ) , h2 = K1 K 2κ 2−0.5 (2) where H denotes the thermal resistance of the contact between roll and strip. Since the radius of curvature of the roll is large compared to the size of the heat affected zone, it is valid to model the roll in a plane configuration. According to FISCHER et al. [17], the dimensionless temperature profile can be described using the dimensionless coordinates ξ and ζ . The circumferential coordinate is given by x=bξ and the radial coordinate is z=ζδ with the heat penetration depth bκ 2 (6) , δ= v where v, b and κ2 denote the rolling speed, the contact length in the rolling gap and the thermal diffusivity of the roll. The dimensionless temperature profile then is ⎛ ζ2 ~ Θ(ξ , ζ ) = ξ exp⎜⎜ − ⎝ 4ξ ⎛ ζ ⎞ π ⎟⎟ − ζ erfc⎜ ⎜2 ξ ⎠ 2 ⎝ ⎞ ⎟. ⎟ ⎠ (3) In the case of heating the roll in the rolling gap the dimensionless temperature of the roll is ~ (4) Θ(ξ , ζ ) = Θ(ξ , ζ ) for 0 ≤ ξ ≤ 1, whereas in the cooling zone the corresponding temperature is given by ~ ~ (5) Θ(ξ , ζ ) = Θ(ξ , ζ ) − Θ(ξ − 1, ζ ) for ξ ≥ 1. A multiplication of the dimensionless temperature function Θ with the maximum temperature, Troll, yields the temperature distribution. The resulting stresses are calculated via constitutive laws [12]. In the linear elastic case the tangential and axial stress components can be estimated by [18] Eα ∆T (7) σ =− , 1−ν where E, ν and α denote Young’s modulus, Poisson’s ratio and the coefficient of thermal expansion, respectively. Plastic deformations are displayed with a kinematic hardening law according to LEMAITRE and CHABOCHE [19]. The flow criterion is (8) σ − q − Y (T ,η) = 0, where q is the backstress and Y is the yield strength as a function of the temperature T and loading direction η (tension: 1, compression: -1). In the following, typical rolling mill values are taken for kinematical, geometrical and thermo-mechanical data [5]. Material data for a HiCr roll are taken from [5, 12, 14, 15]. Figure 10 shows the calculated temperature profile of the roll. Heating in the rolling gap causes a strong increase of the temperature within a thin surface layer of the roll. No noticeable temperature rise can be observed for depths greater than 3 mm. The analytically obtained temperature profile at the exit of the rolling gap and the narrow zone of thermally induced yielding can be seen in Figure 11. Thermal strains cause thermally induced yielding in layers as thick as 180 µm. Deeper into the roll, the strains are purely elastic. 130 INTERNATIONAL DOCTORAL SEMINAR, May 16 – 19, 2010, Smolenice castle Fig. 10 Analytically obtained temperature profile. Fig. 11 Analytically obtained temperature profile at the exit of the rolling gap. Depth range of thermally induced yielding of the work roll material. DISCUSSION In the following the investigated damage of the microstructure is compared with the analytical predictions. It is established that banding is mainly triggered by cyclic thermal loading [5] and that it occurs in areas with a high density of fractured carbides. Therefore, the individual phases of the microstructure and the mechanisms of banding on the micro level are investigated. Figure 12 shows the stress-strain curve calculated from eqs.(1)-(8) for one heating / cooling cycle together with a schematic representation of crack nucleation and propagation. It can be seen that sufficiently high compressive loads in circumferential direction induce short cracks parallel to the roll surface within carbides. Tensile loads in circumferential direction lead to cracks perpendicular to the roll surface within the matrix [5]. Even though the yield strength for tensile loading is not reached in this example, kinematic hardening (Bauschingereffect) of the material may lead to a decreasing yield strength which may be surpassed during subsequent cycles. Through several repetitions of this cycle a crack network is formed. It is assumed that the damaged layer is delaminated by adhesion between the strip and the work roll or the back-up roll and the work roll. This assumption is supported by the experimentally observed transition regions between banded and un-banded areas, compare Figure 4. If the macroscopical mechanical properties of the work roll material are compared with those of its constituting phases (matrix, carbide inclusions), it can be found that the matrix starts to flow during the heat input in the rolling gap. Carbides can be regarded as elastic until their brittle failure [5]. Figure 13 illustrates the shear stress acting on the carbide causing its fracture: Due to the heat input in the rolling gap the material starts to flow plastically. The movement of the material is impeded by the geometry of the roll and the small heat affected layer. Only an expansion in radial direction is possible, whereby the matrix material starts to flow around the carbide. The stresses transferred into the carbide lead to the fracture. Figure 5 shows carbides fractured parallel to the roll surface, which confirms the assumptions made above. All microscopic observations in banded areas show the occurrence of fractured carbides and fire cracks up to depths of 150 µm. This correlates well with the analytically calculated thermally induced yielding zone of 180 µm, see Figure 10. Figure 6 shows surface deterioration with carbides broken out in the banded zone. This corresponds to decohesion of a pre-damaged layer due to adhesive contact in the rolling gap. 131 INTERNATIONAL DOCTORAL SEMINAR, May 16 – 19, 2010, Smolenice castle Figures 7 and 8 show coalesced crack networks. Here, carbides broken perpendicularly to the surface can be seen. Figure 8 shows the transition zone between banded and un-banded areas. A strong deterioration of the transition zone can be observed. Cracks penetrate into the material up to a depth of 150 µm. Fig. 12 Stress-strain curve of one heating / cooling cycle. Depiction of crucial loading states inducing cracks Fig. 13 Simplified sketch of the shear stress inducing fracture of a carbide CONCLUSIONS AND OUTLOOK An analytical model is set up in order to estimate strains and residual stresses in circumferential direction near the surface of a work roll and to correlate these with microscopical observations. Carbides mainly show cracks parallel to the roll surface, caused by a compressive load due to the high temperature gradient in the rolling gap. Crack propagation normal to the work roll surface is assumed to be driven by thermo-mechanical loads induced by high temperature gradients in surface near layers during cooling of the roll. Distributed short cracks of carbides and their coalescence with perpendicular cracks in the matrix (fire cracks) combined with decohesion in the rolling gap are the main failure mechanisms. Fire cracks in areas without broken carbides are not critical regarding banding damage. It can be observed additionally that the maximum thickness of the worn layers does not exceed 150 µm, which corresponds with the extent of the analytically calculated plastic zone. This work complements the efforts of DEBARBADILLO et al. [2] and ARNCKEN et al. [4] and can be seen as an extension with respect to the damage mechanisms on the microscale. According to ARNCKEN et al. [4] a reduction of banding during the hot rolling process could be achieved through water cooling of the strip, shortly before it enters the rolling gap. DÜNCKELMEYER et al. [12] propose a well-considered selection of the rolling speed in order to reduce banding. On the basis of this work guidelines can be set up to reduce banding from a microstructural point of view: Small, equiaxed and statistically uniformly distributed carbides (as is the case in HSS rolls) could reduce the banding tendency of HiCr work roll. However, currently this realisation is hardly possible due to the established production process of the HiCr work rolls. 132 INTERNATIONAL DOCTORAL SEMINAR, May 16 – 19, 2010, Smolenice castle ACKNOWLEDGMENTS The authors gratefully acknowledge the financial support of the Christian Doppler Research Society (CDG). REFERENCES [1] Roll Failures Manual: Hot Mill Cast Work Rolls. The European Foundry Association, Roll Section, 1st edition 2002 [2] J.J. deBarbadillo, C.J. Trozzi, Mechanisms of surface deterioration of hot strip rolling. Iron and Steel Eng., 1981, 63-72 [3] G.Walmag, R.J. Skoczynski and J.P. Breyer, Improvement of the work roll performance on the 2050 mm hot strip mill at Iscor Vanderbijilpark, La Revue de Metallurgie-CIT 2001, 2001, Session 12, 295-304 [4] G. Arncken, T. Heidelauf, M. Möhlenkamp, W. Wolpert, Banding of workrolls in hot strip mills, Technischer Bericht, Thyssen Stahl AG, 1993 [5] T. Reip, Thermomechanische Schädigung von Arbeitswalzen in Warmbreitbandstraßen, Technische Universität München, Lehrstuhl für Werkstoffkunde und Werkstoffmechanik, 2006 [6] L. C. Erickson and S. Hogmark, Analysis of banded hot rolling rolls, Wear Journal, Elsevier Sequoia, 1992, Volume 165, 231-235 [7] N. Narita, K. Yasuda, T. Saitoh, S. Yasunai, Unsteady state temperature analysis of rolls and strips during hot rolling using a simplified comput. algorithm, Proceed. of the 2nd Internat. Conf. on Mod. of Metal Rolling Proc., London 1996, 522-529 [8] A. A. Tseng, Thermal modelling of roll and strip interface in rolling processes, Numerical Heat Transfer, 1999, Volume 35, 135-155 [9] A. A. Tseng, A numerical heat transfer analysis of strip rolling, ASME Journal of Heat Transfer, 1984, Volume 106, 512-517 [10] S. Serajzadeh and F. Mucciardi, Modelling the work-roll temperature variation at unsteady state condition, Modelling and Simulation Materials Science and Engineering, Institute of Physics Publishing 2003, Volume 11, 179-194 [11] C. G. Sun, C. S. Yun, J. S. Chung and S.M. Hwang, Investigation of thermomechanical Behaviour of a Work Roll and of Roll Life in Hot Strip Rolling, Metallurgical and materials transactions A, V29A, 1998, 2407-2424 [12] M. Dünckelmeyer, C. Krempaszky, E. Werner, K. Schörkhuber, G. Hein, Analytical modelling of thermo-mechanically induced residual stresses of work rolls during hot rolling, Steel Research International, Metal Forming 2010, in print [13] T. Reip, C. Krempaszky, L. Renhof and E. Werner, Oberflächenschädigung von Walzenkörpern, Mat.-wiss. u. Werkstofftechnik 2004, 35, No 10/11 WILEY-VCH Verlag [14] T. Reip, C. Krempaszky, E. Parteder, H. Sondermann and E. Werner, Correlation between the Loading of Work-rolls, Banding and Rolled-in Scale, R. Webber, P.C. Perry (Ed.) Proceedings of Material Science and Technology, 2005, Roll Technology, ASM international, 25-29 [15] W. Zhang, C. Krempaszky, E. Werner, Low cycle fatigue behaviour of high chromium iron used for work-rolls, Proceedings of The Seventh International Congress on Thermal Stresses, TS2007 4-7 June, National Taiwan University of Science and Technology, Taipei, Taiwan [16] H. S. Carslaw and J. C. Jaeger, Conduction of heat in solids, Clarendon Press, Oxford, Second Edition 1959, 87-89 133 INTERNATIONAL DOCTORAL SEMINAR, May 16 – 19, 2010, Smolenice castle [17] F. D. Fischer, W. E.Schreiner, E. A. Werner and C. G.Sun, The temperature and stress fields developing in rolls during hot rolling, Journal of Materials Processing Technology 150, 2004, 263-269 [18] H. Parkus, Thermoelasticity, Blaisdell Publishing, Waltham, 1968. [19] J. Lemaitre, J.-L. Chaboche, Mechanics of solid materials, Cambridge University Press, English Edition, Reprint 2000. [20] W. Zhang, M. Dünckelmeyer, C. Krempaszky, Chemical composition and thermo-physical parameter of work-roll materials, Research Report of the Christian Doppler Laboratory of Material Mechanics of High Performance Alloys, Quality related aspects in hot rolling of sheets for automotive applications, August 2008. 134