| Advanced high strength steels (AHSS) deliver greater ductility and formability for equivalent strength levels compared to conventional steels such as mild and high-strength (i.e. HSLA) grades. The combination of higher strength and better formability gives greater energy absorption capabilities that improve crashworthiness for this type of high strength steels. |
 |
| During manufacturing processes, metal may be subjected to a complex history of the strain due to changes in the deformation mode experienced by the material. Therefore, changes in strain path represent one of the most important processing parameters that characterise hot metal forming processes. It has been already proven for different materials that strain reversal affects the onset of both dynamic and static recrystallisation. In most cases, the “yield stress plateau” is observed just after strain reversal which has been attributed to the partial dissolution of the substructure during the first stages after the deformation mode change. Also, due to that change, the kinetics of both dynamic and static recrystallisation are retarded. There is however a lack of information concerning the deformation products of those materials when they are cooled down to room temperature. |
 |
| Residual stresses are well known to affect the fatigue, fracture and creep failure of metals; tensile stresses can contribute to driving crack growth whilst compressive stresses are inhibitive. Compressive surface stresses have been shown to influence toughness and crack path followed during fracture. Controlled plasticity burnishing allows concentrated regions of surface residual stress to be generated with a high degree of spatial resolution. This technique was applied to modified double cantilever beam specimens (mDCB) and compact tension (CT) specimens to near 1D residual stress fields. Investigations were conducted into the fracture behaviour of these specimens with discrete burnished regions; CTOA measured using direct techniques has been used to characterise fracture toughness and observe its variation with stress field. |
 |
| Crystal plasticity finite element models have been developed to simulate deformation microstructures during metals processing. As the first application, the uniaxial tension of a copper single crystal has been simulated and results have been compared to other models found in the literature [1] showing good agreement. The plane strain compression of an aluminium alloy has then been investigated using a large number of grains in three dimensions. The rotation of individual grains during deformation can be predicted successfully using the model. The predicted {100} and {111} pole figures of the deformed microstructure are indeed very similar to experimental results (Figure 1). Finally, this model has been used to simulate the local deformation of polycrystalline AA5052 during a plane strain compression test with embedded microgrids. |
 |
| The high shear processing (HSP) during hot rolling of aluminium is particularly effective in producing a highly deformed subsurface layer, due to asperity contact between stock and work rolls, which enhances filiform corrosion. Another example is modelling of heat generation and flow during friction stir welding (FSW) process, which can be considered as the basis of all other models of the process, be these microstructural, computational fluid dynamics or thermomechanical. Understanding and prediction of physical phenomena, which are taking place at the tool/workpiece interface during HSP, becomes very difficult using the traditional finite element (FE) techniques. The potential of FE tools and techniques merged with discrete element (DE) based transient dynamics is highlighted. |
 |
|
Micromechanical modelling of damage and fracture makes it possible to assess the integrity of structure not only in small scale yielding conditions but also in conditions where large scale plasticity exists. Compared to conventional fracture mechanics such an approach is independent of the geometry and size of the structure as long as the same type of fracture is expected in the structure. Application of the local approach, consists of identification of micromechanical behaviour of material and implementation of suitable constitutive equations and damage models capable of capturing the stress/strain coupling with damage evolution. These damage models need to be tuned by performing correlations of numerical simulations and physical parameters on laboratory test specimens. However, lack of a systematic approach to tune the damage models is observed in the literature.
|
 |
|
Friction Stir Welding (FSW) is a new technique, being marketed to industry, which utilizes frictional heating combined with forging pressure to produce high strength bonds virtually free of defects. Friction Stir Welding transforms metals from a solid state into a “plastic like” state, and then mechanically stirs the materials together under pressure to form a welded joint. Although considerable experimental work has been reported in the literature, the process simulation work appears to be at a relatively early stage.
|
 |
|
The production of high integrity forgings particularly on the nuclear sector is gathering momentum. The development of high purity SA508 materials together with new advances on numerical analysis offers the possibility of integrating both modelling and metallurgy on the design cycle for such components. Preliminary work at Sheffield Forgemasters on high purity SA508 alloy suggests that ideal mechanical properties needed for nuclear applications can be obtained. To ensure that the suitable material properties are obtained, finite element analysis will be used to look at the material flow, heat treatment and tool design. Finally, the above will be integrated to produce a forging prototype for further characterization analysis. At this stage in the project one large scale forging has been produced, showing outstanding results in comparison to the model.
|
 |
|
The granular computing neural-fuzzy (GrC-NF) framework is investigated in this project from a modelling robustness perspective. This framework has been used in the past and it is shown that it can provide models with very good accuracy, transparency and generalisation properties. This project investigates how such models manage with the inaccuracies associated with real industrial data such as measurement scatter, errors and variability as well as outlier points (see Figure 1). This work shows how such information (error, variability) can be used by the modelling methodology in order to improve the prediction performance unlike most outlier/noise detection algorithms that aim to detect and eliminate (delete) such data points. Problematic data still contain useful information that the proposed algorithm aims to use to its advantage rather than deleting them from the data base. Using a benchmark data set it is shown how this methodology is applied to use error tolerance data in order to improve the system’s performance (see Figure 2). Additionally, the granulation algorithm is expanded to include a methodology that adjusts the granular merging of outlier data.
|
 |
|
In classical data driven modelling, such as statistical regression or neural-fuzzy modelling, it is often assumed (explicitly or implicitly) that the model error is random and follows a normal distribution. This assumption is frequently invalid in many modelling scenarios, e.g., when there exist different random disturbances or there are large measurement scatters (such as is the case of Charpy impact energy data modelling).
In this research an expectation maximisation (EM) algorithm is proposed to extract the probabilistic characteristics from the model error, under the assumption of Gaussian Mixture Model (GMM). |
 |
|
Transformation of the high temperature ? phase during cooling in titanium alloys can be complex. Depending on cooling rate, decomposition may involve diffusion controlled nucleation and growth, martensitic and possibly massive mechanisms [1, 2]. The ??? phase transformation generally is governed by the Burgers orientation relation [3]:
(0001)? // {110}? , <1 10>? // <111>?
Due to crystal symmetry, there are 12 possible ? orientations, or variants, that can result from the transformation of a single ? parent grain. Under certain conditions some variants may occur more frequently than others, a phenomenon known as variant selection |
 |
|
Recent advances in computing power have led to significant advances in experimental techniques available for material characterisation especially in terms of measurement. A technique of measurement which has made significant progress is digital image correlation and is now the preferred choice for full field displacement measurements. At the Dept of Mechanical Engineering, University of Sheffield, this technique has been used extensively for material characterisation and is currently being extended for use in evaluation of structures. Recent acquisition of a 3D system has further extended the capability of the technique. This presentation will highlight results obtained from a range of different experiments (eg. tensile, uniaxial ratchetting and tear tests). The resolution of the full field measurements have revealed phenomenon which have been previously been missed or ignored when using other more conventional measurements techniques. Preliminary results from structural evaluations have shown very small deformations (approximately 45 microns) in a 100kN mechanical test frame at full capacity.
|
 |
|
|
|