ENHANCING THE LIFE OF SHREDDER HAMMER TUNGSTEN CARBIDE TIPS By J.G. LOUGHRAN 1, S.I. ANDERSON 2, J. CAMUGLIA 3 and N. TRAPP 3 1 School of Engineering, James Cook University 2 Rockfield Technologies Australia Pty Ltd 3 Boogan Implement Company Pty Ltd KEYWORDS: Shredder-Hammer, Life, Damage, Comminution, FEA. Jeffrey.Loughran@jcu.edu.au Abstract THIS paper reports on a computational and experimental investigation into the dynamic impact of steel debris against sugar mill shredder hammer tips. During cane preparation, billets of cane are comminuted against tungsten carbide (WC) tips with impact velocities approaching one-third the speed of sound. Foreign debris (often hardened steel pins) in the input stream, result in a high probability of fracture to the WC tiles. During fabrication, the WC tiles are brazed to a mild steel or white iron backing-block using a tri-ply laminate. The laminate is several orders of magnitude less stiff than the WC. Besides acting as a fastener the laminate serves to minimise the build-up of residual stress in the WC due to the marked difference in coefficient of thermal expansion between adjacent materials. The laminate also influences the impedance as compression waves are reflected and refracted. The impact process when steel debris comes into contact with the WC is modelled using explicit finite element dynamics. A Rankine rotating crack model is used to characterise the WC and the projectile is modelled as elasto-plastic. The effect of the WC geometry on resistance to crack initiation is noted and explored numerically. The mechanisms, which dominate failure, are reported. Laboratory experiments confirm the computational modelling. Factory observations involving novel tip geometries at Tully Mill provide evidence of throughputs exceeding 750 000 tonnes of cane between replacements. Introduction Many industrial processes involve the impact of foreign debris against quasi-brittle materials. The subsequent damage can be erosion or in extreme cases catastrophic failure. In the sugar processing industry, for example, foreign debris such as hardened steel pins often enter the cane supply to the factory. The cane is comminuted in a heavy duty swing hammer shredder (Figure 1). The input stream consists of billet cane and other foreign debris (dirt, 508
rocks, tramp iron etc.). This stream is struck by a hammer assembly with a tip speed of around 100 m/s. The billet cane is immediately pulverised into pith cells and damaged fibrovascular bundles ranging in size from a few millimetres to 60 mm in length by 2.5 mm in cross-section. To minimise wear due to dirt in the input stream, WC is brazed to a mild steel backing block at the tip of the hammer. This is a very effective solution provided other foreign debris such as steel can be excluded from the cane supply. The latter is impossible given the high throughputs of modern factories, and there is always a trade-off between increasing the hardness of the WC to minimise wear and reducing the hardness to improve fracture resistance. Several authors have reported to the ASSCT on shredder tip investigations starting with Mason et al. (1979), Dolman (1983), Lakeland et al. (1992) and Ostlund et al. (1996). Mason reported on experiments using an SRI experimental shredder. Lakeland investigated tips from a metallurgical standpoint. Ostlund reported on preliminary finite element modelling of the transient dynamic impact process. This paper builds on the earlier work of Ostlund (1996) using the non linear finite element software ELFEN. Particular attention is focused on the effect of geometry and contact location in controlling the evolution of damage that occurs when the WC material is struck with foreign debris. (a) (b) Fig. 1 Schematic of cane shredding equipment: a) Cross-sectional sketch of a heavy-duty swing hammer shredder showing input and output streams; b) A typical shredder hammer assembly consisting of a tungsten carbide tile (upper right) brazed to a mild steel backing block. This assembly (called the tip) is mechanically fastened to the hammer. On occasion, the mild steel backing block may incorporate a layer of white iron to give additional wear resistance in case of failure of the WC tip. Manufacturing process for the WC tip During tip manufacturing, the WC tile is brazed to a mild steel backing block using a tri-ply Ag-Cu-Ag shim. The assembly is heated to around 700 C at which time the Ag wets adjacent surfaces. During the cooling process, the Cu yields and bending stresses develop in the WC due to significant differentials in the coefficient of thermal expansion between adjacent materials (about a factor of three). The effect of the residual thermal stress was computed by Ostlund et al. (1996). It is noted that bending stresses at the top of the tile from 509
the brazing process can be as high as 200 MPa. In this paper, we ignore the residual stress and concentrate on modelling of the impact process. The potential effect of the residual stress is discussed in concluding remarks. Constitutive law for large strain deformation of brittle materials The WC is a quasi brittle material with little ductility when subjected to impact. To model multifracturing phenomena, existing strategies range from continuum based finite element procedures involving cohesive zone models and discontinuous Galerkin formulations (Borst, 2001; Ruiz et al., 2001), to discontinuous driven formulations (Shi, 1988) and distinct/discrete analyses (Cundall and Strack, 1979). More recently, we have seen evidence of combined finite/discrete procedures (Owen et al., 2003; Reynolds and Loughran, 2004). The latter is particularly powerful for modelling of ballistics/blast applications where extreme damage may result in an evolutionary increase of 3 4 orders of magnitude in deformable but discrete bodies. Although such fragmentation does not occur in the present application, the underlying physics associated with the continuum to discrete fracture procedures appears ideally suited to the current problem. The coalescence and development of microcracks within the continuum is assumed to occur in directions that attempt to maximise the energy release rate and simultaneously minimise the strain energy density. The manifestation of a discrete fracture results in the localisation of inelastic strains and consequential unloading of surrounding material. Under transient loading, these macroscopic fractures tend to align with the direction of maximum principal strain. Subsequent localisation of the micro-cracking into crack bands then results in the softening process occurring orthogonal to the principal crack directions. The constitutive model employed is an anisotropic Rankine rotating crack plasticity model, described in detail by Cottrell et al. (2003). The model exhibits a Mohr Coulomb elasto-plastic response in compression with a hydrostatic cut-off in tension. Damage occurs at a spatial level when the ultimate tensile strength of the material is exceeded and the local stiffness is reduced to zero. Using this procedure, a continuum with micro-cracks can be treated as an equivalent anisotropic material with degraded properties orthogonal to the crack surface. A rotating crack model is then used to ensure alignment of cracks with the principal axes. Finite element modelling Brazing analysis The thermal analysis undertaken by Ostlund et al. (1996) using ABAQUS was remodelled here with ELFEN. Material property data are presented in Table 1. Using a plane 510
strain model for the tip and a fully non linear semi-coupled therma /structural analysis, stress results were obtained (Figure 2) which agrees largely with the earlier predictions. It can be seen that the tri-ply acts as a thin plastic fastener connecting the relatively stiff tungsten to the more flexible backing material. A maximum tensile stress of around 200 MPa occurs at the top centre of the WC. In contrast, the centre of the base of the WC experiences a compressive stress of around 60 MPa. Table 1 Mechanical properties of hammer tip components for thermal analysis. Material Elastic modulus E (GPa) Yield strength (MPa) Tensile strength UTS@ε uts (MPa) Coefficient of thermal expansion Mild steel 207 250 400@0.15 14.1E-6 White iron 207 250 400@0.15 14.1E-6 Brazing 73 80 220@0.15 19.9E-6 WC (H10P) 558 N/A N/A 5.3E-6 Fig. 2 Results of thermal analysis: Stress contour plot (left); Deformed response magnified by 10 (right). Note the elongation in the Cu laminate. Impact analysis Modelling of the impact of steel projectiles against the WC tip requires a fully non linear transient dynamic analysis. Maximum contact energy between the projectile and WC tile will be realised when the projectile strikes under plane strain conditions. Hence, a plane strain analysis is appropriate. A typical hammer has a mass of about 20 kg compared to a tip mass of about 3.5 kg. The centripetal loading is significant (equivalent to about 140 kn). The choice of fixed boundary constraints around the base of the steel assembly is a reasonable engineering approximation. The impact analysis was conducted assuming no residual stresses from the brazing process. The effect of the residual stresses is considered via engineering interpretation under concluding remarks. Figure 3 shows a schematic of the modelling set-up. 511
Fig. 3 Schematic describing the plane strain model set-up for analysis. The elasto-plastic projectile (arrowed) strikes the quasi brittle WC tile (thick, dark area) at an assumed tip velocity of 100 m/s. The WC tile is fastened to the mild steel backing block (grey) by the tri-ply (thin dark strip). In practice, a wide assortment of steel debris enters the shredding system and spatial contact with the WC tile will be stochastic. For simplicity, the projectile is modelled as high strength steel. Two primary impact positions were considered a centre impact and an edge impact. Dimensional parameterisation was applied to the projectile so that models could be rerun to assess the effect of projectile size on damage. Table 2 lists pertinent mechanical property data. Typical mesh discretisation is given in Figure 4. Because the problem is geometrically and materially nonlinear and transient, linear triangular elements are employed within an explicit computational environment. Table 2 Pertinent material data required for finite strain numerical modelling. Material Elastic modulus E (GPa) Yield strength (MPa) Tensile strength UTS @ε uts (MPa) Fracture K IC (MN/mm 3/2 ) Hardness HV30 Mild steel 207 250 400@0.15 Projectile 207 827 1034@0.15 Tri-ply 73 80 220@0.15 WC (H10P) 558 1300 17 1210 512
Fig. 4 Finite element mesh for modelling impact. Figure 5 shows an exaggerated view of the bending stress associated with the two target positions. The centre impact results in bending of the tile with maximum tensile stress occurring at the base of the tile. In contrast the edge impact results in a maximum tensile stress at the top surface. The evolution of damage for both target positions is depicted in Figure 6. Fig. 5 Bending failure modes (exaggerated). 513
A simple numerical experiment was performed to assess the diameter of the projectile to instigate fracture at each target location (Figures 6 and 7). The results showed that an 11.5 mm diameter rod would initiate fracture for a centre target. In comparison, an edge impact required a 9.6 mm diameter rod. Improvements to the tile design can be achieved by three means: (1) geometric changes, (2) material changes and (3) a combination of (1) and (2). Geometric changes are considered here. Improving the centre bending performance can be achieved by increasing the moment of inertia of the section. One option is simply to thicken the WC tile. However, the additional material at the edges will have negligible effect as the critical region is at the centre. Likewise, a thicker WC tile will increase the moment of inertia and minimise damage for an edge impact. In this case, the thicker section is only required for half the tile towards the free edge. A consequence of selecting a thicker tile is that at some point the bending type failure mechanism no longer applies and a new failure mechanism is activated. This switch is illustrated by considering an Edge failure. At a certain tile thickness, the localised stress close to the projectile exceeds the failure stress before the failure bending stress can be reached. A parameter study was conducted on the geometry to approach a near optimal solution. The best geometry is depicted in Figure 8. Experimental investigation Laboratory testing A high speed pneumatic projectile firing device was manufactured and installed at the BIC factory. The facility was capable of firing hardened steel pins of 25.4 mm diameter x 90 mm in length at speeds of up to 100 m/s. The shredder hammer tips were rigidly located inside an enclosed safety chamber. Figure 10 shows a schematic of the facility. Test results which are in general agreement with the numerical modelling are also included with Figure 9. Factory results New tips were installed in the Tully shredder for the 2005 crushing season. Figure 10 shows a photograph of two of the tips which were removed during a regular maintenance period. The tip at the top of the photograph had processed 631 723 tonnes while the one at the bottom, 401 816 tonne. The general appearance of both tips is good and it is mooted that both tips could have processed further tonnage. 514
(a) Principal stress prior to fracture [Pa]. (b) Crack initiation. (c) Complete failure. Fig. 6 Centre-bending failure (diameter = 11.5 mm). 515
(a) Principal stress prior to fracture [Pa] (b) Crack initiation (c) Complete failure Fig. 7 Edge bending failure (diameter = 9.6 mm). 516
Fig. 8 Optimised WC tile geometry to minimise damage under dynamic impact with hardened steel pins. Fig. 9 Experimental laboratory testing: High speed projectile facility at BIC (Left); Evidence of bending failures when a hardened steel projectile (20 mm diameter x 50 mm long) impacted tiles end on at 100 m/s (right). Fig. 10 Photograph showing wear characteristics of WC tips from Tully Mill 2005: 631 723 tonnes (top); 401 816 tonnes (bottom). 517
Conclusions An explicit transient finite element analysis has been conducted to explore the evolution of damage that occurs when high strength steel impacts a comminution tool manufactured from tungsten carbide. A Mohr Coulomb rotating crack constitutive model was employed in a finite element framework where mesh topology was updated as cracks initiated and propagate. A parametric study of tile geometry was conducted to develop an improved understanding of the dominant fracture mechanisms. An improved tile geometry was devised and tested both in the laboratory and factory. The new tile geometry has improved resistance to fracture from impact with tramp iron. In a factory trial at Tully in 2005 over 600 000 tonnes of cane was shredded and it was mooted that the tile could have achieved an even greater tonnage. REFERENCES Borst, R.D. (2001). Some recent issues in computational failure mechanisms. Int. J. Numer. Meth. Engng., 52 (5): 63 96. Cottrell, M., Yu, J., Wei, Z.J. and Owen, D.R.J. (2003). The numerical modelling of ceramics subject to impact using adaptive discrete element techniques. Engineering Computations, 20 (1): 82 106. Cundall, P.A. and Strack, O.D.L. (1979). A discrete numerical model for granular assemblies. Geotechnique, 29: 47 65. Dolman, K.F. (1983). Alloy development: Shredder hammer tips. Proc. Aust. Soc. Sugar Cane Technol., 5: 281 287. Lakeland, K.D., Loughran, J.G. and Gay, S. (1992). Experiments with cane shredder tip materials. Proc. Aust. Soc. Sugar Cane Technol., 14: 281 287. Mason, V., Perrott, C.M. and Sare, I.R. (1979). Wear resistant materials for sugarcane shredder hammers. Int. Conf. on Wear of Materials, Dearborn, USA. Owen, D.R.J., Andrade Pires, F.M., De Souza Neto, E.A. and Cottrell, M.G. (2003). A hybrid continuous/discrete representation of fracturing solids. Proceedings of IMECE, 03: 1 25. Ostlund, S., Loughran, J.G. and Meyers, T. (1996). A preliminary investigation into the mechanics of tramp iron impacting on tungsten shredder hammer tips. Proc. Aust. Soc. Sugar Cane Technol. 18: 292 297. Reynolds, T. and Loughran, J.G. (2004). Modelling penetration mechanics of composite confined ceramic targets subject to high speed projectile impact. 21 st International Symposium on Ballistics, Adelaide. Ruiz, G., Pandolfi, A. and Ortiz, M. (2001). Three-dimensional cohesive modelling of dynamic mixed mode fracture. Int. J. Numer. Meth. Engng., 52 (5): 97 111. Shi, G. (1988). Discontinuous deformation analysis: a new method for computing stress, strain and sliding of block systems. PhD thesis, University of California, Berkeley. 518