Rajesh Ransing

Powder compaction modelling via the discrete and finite element method

Abstract A discrete and continuum modelling approach to powder compaction is presented and discussed. The work demonstrates the ability of the discrete method to capture the compression of a ductile matrix and a matrix comprising a ductile and brittle assembly. The discrete element analysis is compared with a Gurson model for the compression of a ductile porous system with reasonable agreement. A continuum model is demonstrated by comparison with results from an experimental study and other numerical investigations. For a steel powder, this displays agreement in predicted density within 0.23 g/cm 3 and upper punch force level within 16%. These levels of agreement are encouraging and are within the requirements to be useful for industrial application. Similar good agreement is also demonstrated when compared with a simulation of the pressing of a ceramic tip which features extreme changes in density throughout the compact.

Numerical comparison of a deformable discrete element model and an equivalent continuum analysis for the compaction of ductile porous material

Abstract The combined finite and discrete element technique has been considered for the compaction of an assembly of particles. Each particle is mapped with finite elements and the interaction between particles is solved using a discrete element technique. Compaction of a few particles in two dimensions using the combined discrete/finite element method is shown to be equivalent to the Gurson model in the continuum analysis. Results generated from both analyses are compared and found to be in good agreement up to 95% densification. The work demonstrates the potential applicability of the deformable discrete element method for particulate analysis or powder compaction in particular.

Application of neural computing in basic oxygen steelmaking

Abstract The use of computer control in the basic oxygen steelmaking (BOS) process is essential to obtain accurate end-point temperature (EPT) and carbon control in liquid steel. The current computer model employed to execute this task is a procedural model that must be maintained by a person with considerable steelmaking knowledge. The requirement for an improved, maintenance reduced model is becoming increasingly important as expertise in this area is dwindling. The steelmaking process is highly complex and volatile. Artificial neural networks (ANNs) have been used to model this type of non-linear system. This paper describes an investigation into the use of ANNs to predict oxygen and coolant requirements during the end-blow period of the steelmaking process. During the end-blow period, a temperature measurement and sample are taken using a probe. These measurements are then used as inputs to the ANN model in order to predict how much oxygen to blow and how much coolant to add in order to achieve the desired end-point conditions in the steel at the end of the process. The software used to perform most of the modelling was the Clementine Data Mining System. This paper discusses the results from the ANN trials at Port Talbot BOS plant, which is part of the Corus Group.

A discrete deformable element approach for the compaction of powder systems

The paper describes the application of a discrete deformable element modelling approach to simulate the compaction of a mixture of ductile and brittle powders together with an exploration of its suitability to establish the yield characteristics of powders. Two-dimensional rod models comprising ductile and brittle particles are assembled and subject to uniaxial compression in a rigid die. The utility of the method to represent the yielding of the powder is also explored by the application of a biaxial loading sequence. The model was found to be capable of modelling the behaviour of powder mixtures successfully. It captured ductile–brittle failure mechanisms in the case of brittle particles and demonstrated the effectiveness of a proportion of ductile particles in preventing the fragmentation of brittle particles in a mixture, illustrating agreement with experimental observation. The exploration of yield surface prediction was subject to a number of assumptions driven by the two-dimensional nature of the simulation, coupled with the high initial density level enforced by the particle packing. Despite these constraints, the results were encouraging when compared with experimental measurement of yield characteristics. Further work is required to establish more realistic initial density and to develop the capability to include a larger number of particles within the simulation.

An Improved Back Propagation Neural Network Algorithm on Classification Problems

The back propagation algorithm is one the most popular algorithms to train feed forward neural networks. However, the convergence of this algorithm is slow, it is mainly because of gradient descent algorithm. Previous research demonstrated that in ‘feed forward’ algorithm, the slope of the activation function is directly influenced by a parameter referred to as ‘gain’. This research proposed an algorithm for improving the performance of the back propagation algorithm by introducing the adaptive gain of the activation function. The gain values change adaptively for each node. The influence of the adaptive gain on the learning ability of a neural network is analysed. Multi layer feed forward neural networks have been assessed. Physical interpretation of the relationship between the gain value and the learning rate and weight values is given. The efficiency of the proposed algorithm is compared with conventional Gradient Descent Method and verified by means of simulation on four classification problems. In learning the patterns, the simulations result demonstrate that the proposed method converged faster on Wisconsin breast cancer with an improvement ratio of nearly 2.8, 1.76 on diabetes problem, 65% better on thyroid data sets and 97% faster on IRIS classification problem. The results clearly show that the proposed algorithm significantly improves the learning speed of the conventional back-propagation algorithm.

The optimal design of interfacial heat transfer coefficients via a thermal stress model

Abstract The thermo-mechanical analysis of a solidification process has been presented to predict an air gap width. The air gap width value has been used to approximate the interfacial heat transfer coefficients. The Lewis-Ransing correlation has been used to link the stress model with an optimization model for the optimal feeding design.

Using a deformable discrete-element technique to model the compaction behaviour of mixed ductile and brittle particulate systems

This paper illustrates the application of a combined discrete– and finite–element simulation to the compaction of assemblies comprising both ductile and brittle particles. Through case studies, the results demonstrate the importance of using a fine mesh on the particle boundary, the effect of fragmentation and its impact on the form of the compression curve, and the effect of inclusion of ductile particles at ca. 25% by volume suppressing brittle failure mechanisms. Although, the calculations can be extended to three dimensions, the computational cost is a current limitation on such calculations. The novelty of this approach is in its ability to predict material yield surfaces for the compaction of a mixture of particles. The initial results are optimistic, but there is a need for model improvement, principally through the ability to capture the random packing of irregular particles since this will eliminate a key problem in defining an initial density for the simulation. The main advantage of this technology is in its ability to minimize the need for expensive triaxial testing of samples to develop the yield–surface history.

Casting shape optimisation via process modelling

Abstract The paper describes the application of numerical optimisation case studies to two practical cast components. The first study highlights the potential benefit of thermal controls to achieve an optimised cast part, whereas, the second focuses on shape optimisation. In comparison with practice, it was noted that the unconstrained optimisation for thermal control identified the improvement path correctly, but the solution could not be applied practically. The shape optimisation study also identified the correct improvement path and the result was supported by trials that had been completed previously in the foundry.

Alternative techniques for casting process simulation

Over the past 20 years, casting process simulation has been an active area of research. The simulation techniques are either based on solving governing partial differential equations using numerical schemes such as the finite element or finite difference methods, or a variety of heuristically based geometry driven methods. Numerical methods are more accurate, but geometry driven methods are computationally less expensive. This paper explores two alternative techniques to overcome some of the limitations of traditional numerical simulation schemes for the casting process simulation. The first technique uses a geometric transformation method known as the medial axis transformation, to predict hot spots whereas the second technique, based on meshless methods, is used for simulating the mould filling process.

Enhanced medial axis interpolation algorithm and its application to hotspot prediction in a mould–casting assembly

Abstract The use of a medial axes transformation technique to predict temperature profile and hot spots in solidifying castings is investigated. In essence, a simplistic yet comprehensive temperature interpolation algorithm is proposed, which can solve the solidification problem qualitatively and quantitatively. The method combines the advantages of geometric reasoning and conventional numerical simulation. Previous work by the authors (valid only for a casting geometry without a mould) is enhanced and modified to take into account the effect of the mould and to consider the interaction of the whole mould–casting assembly. The feasibility of the proposed technique is demonstrated by comparison with both a pure geometric reasoning technique and finite element modelling. In addition, numerical predictions of the model are validated against real-life casting geometries.