The work presented here is part of a broader study concerning the biomechanical analysis of the movement of dinosaurs, which will be done by the finite element method (FEM). For this aim, it will be necessary to count on virtual models of the walking system (that is to say, the foot, leg, etc.) and the substrate on which the dinosaur moved. Both kinds of models can be approximately inferred from fossil remains: bones for the former and ichnites (fossil footprints) for the latter. Obviously, there are important challenges in these models, let us see, for example, that a group of fossil bones (probably incomplete and deteriorate) is very far from a walking animal with not only the bones and the articulations but also with flesh, tendons, skin, claws and so on. In any case, fossil bones are the only material we have to start modelling. Therefore, the first step will be their 3D digitization. As the expected use of the 3D models defines, the technical characteristics that these models need to comply with, the manuscript will reflect on the qualities that the models for biomechanical purposes need, paying attention to the completeness, geometric accuracy and resolution. Moreover, a practical case is presented with a comparison of a scanning technology (fringe projection) and close range photogrammetry in order to model a 2 meters tall leg of a specimen of Edmontosaurus.

One of the main objectives when designing welded products is to reduce strains and deformations. Strains can cause excessive angular distortion. This results in a welded product that does not meet acceptable tolerances. The geometry of the weld bead height and width depends on the input parameters speed, voltage and current, and provides the welded joint with strength and quality. As welded products become increasingly complex, deformations become more difficult to predict as they depend greatly on the welding sequence. This paper shows how a combination of the Finite Element Method, Genetic Algorithms and Regression Trees may be used to design and optimize complex welded products. Initially, Artificial Neural Networks and Regression Trees that are based on heuristic methods and evolutionary algorithms were used in predicting the weld bead geometry according to the input parameters. Then, thermo-mechanical Finite Element models were created to obtain the temperature field and the angular distortion using the weld bead geometry that the best predictive models generated. Finally, optimization techniques that are based on Genetic Algorithms were used to validate these Finite Element models against experimental results, and to subsequently find the optimal welding sequence to use in the manufacture of complex welded products.

Intramedullary (IM) nails are commonly used in the fixation of long bone fractures such as fractures of the femur and tibia. They are commonly hollow, rigid rods made from 316L steel or titanium alloys. Static locked IM nails which ensure that both ends of the long bone do not move relative to each other have traditionally been used, however, recent research suggests that such rigid bone-nail constructs do not provide optimal conditions for bone healing. Dynamic fixation allows the bone fragments to move a prescribed distance relative to each other, thus causing micro-motions at the fracture site and stimulating healing via callus formation. This paper describes the development and proof of concept of a new dynamic IM nail, incorporating an internal mechanism which allows for a tailored micro-motion at the fracture site. In vitro bench testing using synthetic composite bones (Sawbones, USA) was carried out to compare the performance of the new device with standard IM nails under compression, four-point bending and torsional loads. It was found that the dynamic implant performed well in comparison the standard nail and provided the required micro motion at the fracture site.

Welding is a metal joining process widely used in industry. In this process, homogeneous microstructures, residual stresses, and variations in mechanical properties greatly affect the quality of welded joints. However, defects can occur due to intense concentration of heat in the welded region. These defects vary depending on the type of the welded material, welding process, and cooling rate of the welded parts. Over the years, different standard tests have been performed by researchers to measure and to quantify welded joints. Hardness and impact testing have been widely used to evaluate microstructures and variations in mechanical properties of welded joints. Hole drilling and x-ray diffraction have been developed to evaluate the residual stress. Tensile tests are performed to obtain the tensile and yield strengths of welded components. This chapter aims to show and compare the variations of hardness, microstructure, tensile properties, residual stresses, and impact strength for the most important welding processes in welded joints. The work focuses solely on ferritic and martensitic stainless-steel materials, for similar and dissimilar materials.

This paper shows a methodology for adjusting the top-bottom contact stress ratio belonging to the outer raceway by combining the loads acting on the bearing by means the finite element method (FEM) and response surface methodology (RMS). A three-dimensional Finite Element (FE) model considering the material properties and geometry of a real tapered roller bearing was built. Subsequently, a design of experiments (DoE) study varying the input loads (preload, radial load, axial load and torque) was implemented. These input loads generated thought DoE were simulated in the FE model, and the contact stresses belonging to the outer raceway were obtained. A quadratic regression model for predicting the ratio between the contact stresses was formulated, and using the RMS method, a combination of input loads were found for adjusting a specific top-bottom contact stress ratio.

An experience is presented using the finite element method (FEM) and data mining (DM) techniques to develop models that can be used to optimize the skin-pass rolling process based on its operating conditions. A FE model based on a real skin-pass process is built and validated. Based on this model, a group of FE models is simulated with different adjustment parameters and with different materials for the sheet; both variables are chosen from pre-set ranges. From all FE model simulations, a database is generated; this database is made up of the above mentioned adjustment parameters, sheet properties and the variables of the process arising from the simulation of the model. Various types of data mining algorithms are used to develop predictive models for each of the variables of the process. The best predictive models can be used to predict experimentally hard-to-measure variables (internal stresses, internal strains, etc.) which are useful in the optimal design of the process or to be applied in real time control systems of a skin-pass process in-plant.

Tapered roller bearings are mechanical transmission elements capable of supporting axial and radial loads, both under static, dynamic or variables conditions. All these load combinations on the bearing are capable of inducing high contact pressures on rolling surfaces (raceway inner, raceway outer and rollers) and the relative displacements between the different component parts of the bearing. These high contact pressures on the raceway cause phenomena such as pitting, decreasing the durability of mechanical components significantly. Today, the design of this type of bearings is still based on both analytical techniques and experimental techniques, as well as on the finite element method (FEM). This paper explains the process of setting a finite element model (FEM) of a tapered roller bearing mounted on a vehicle´s axle. To adjust this FE model, Non Linear Submodeling techniques were utilized successively.

The aim of this paper is to develop a methodology for the validation of a Finite Element Model (FE model) which represents commercial brake caliper. The materials’ characteristics are totally unknown so Genetic Algorithms techniques and experimental tests such as test temperature and deformations have been used to determine them. Finally it presented a general vision of these techniques’ potential to reduce the costs of testing and prototyping models which have been replaced by Finite Element Method (FEM).

This paper presents a method based on genetic algorithms and the finite element method which is useful for automatically adjusting the parameters of a tension levelling process. First, the optimum parameters of the steel to be used in the simulation programme are sought. The process consists of simulating controlled cyclical deformation tests on finite element (FE) models of standard steel test pieces with different laws of cyclical behaviour. Genetic algorithms are used to optimise the parameters of the simulation model so that the behaviour of the material is as close as possible to the results obtained in real experimental tests. This ensures that the behaviour of the material in the FE model is as realistic as possible. The model of behaviour of the material selected is used to design and check out a second tension levelling FE model. Based on this second model, the roll penetration, the lengthwise tension and the strip feedrate are adjusted. There is also optimisation with genetic algorithms so that the final residual tensions in the product are below a specified threshold and as even as possible. For a solution to be considered as valid, it must be confirmed that the steel plate is subject to the tensions envisaged at various process control points. From the best solutions found, the one with the fastest feedrate is selected so as to maximise output.

To ensure realistic results in modelling processes for analysing strains in material using finite element (FE) models, it is essential to have a model of the material that is as close to reality as possible, especially when materials are subject to cyclic loads, because the gap in behaviour between actual materials and simulated models widens as the number of cycles increases owing to the Bauschinger effect, ratchetting, and other effects. This paper sets out a fully automated method for determining the most appropriate material behaviour model (linear or non-linear) for use in numerical simulation programs and the optimum constitutive parameters that define that model, on the basis of experimental data and the combined use of genetic algorithms (GAs) and finite elements (FEs). As a practical example, the method is applied to determine the optimum material model for ZSTE 800 high-strength steel with a view to simulating its behaviour in a cyclic stress—compression process with controlled strain and a variable number of cycles.

This paper describes a method based on a combination of Finite Element Method (FEM) and Data Mining (DM) techniques to set up prediction models that can be used to calculate bolted connections. Based on the results of a finite element (FE) model validated by tests, a number of FE simulations is developed, varying the most significant parameters (thickness, bolt diameter, friction, etc.). The results of these simulations are used to generate a database which can then be used to create prediction models. The process centres on selecting the best technique from a set of DM and artificial intelligence (AI) algorithms to find the models which provide the most generally applicable solutions to the problem.This method, combining FE models with prediction techniques, is highly useful for the specific case of bolted connections, because it enables results to be obtained almost in real time with only slight prediction errors. This makes it an excellent tool for optimising the design of such connections. (C) 2010 Elsevier Ltd. All rights reserved.

This paper demonstrates that combining regression trees with the finite element method (FEM) may be a good strategy for modelling highly non-linear mechanical systems. Regression trees make it possible to model FEM-based non-linear maps for fields of stresses, velocities, temperatures, etc., more simply and effectively than other techniques more widely used at present, such as artificial neural networks (ANNs), support vector machines (SVMs), regression techniques, etc. These techniques, taken from Machine Learning, divide the instance space and generate trees formed by submodels, each adjusted to one of the data groups obtained from that division. This local adjustment allows good models to be developed when the data are very heterogeneous, the density is very irregular, and the number of examples is limited. As a practical example, the results obtained by applying these techniques to the analysis of a vehicle axle, which includes a preloaded bearing and a wheel, with multiple contacts between components, are shown. Using the data obtained with FEM simulations, a regression model is generated that makes it possible to predict the contact pressures at any point on the axle and for any condition of load on the wheel, preload on the bearing, or coefficient of friction. The final results are compared with other classical linear and non-linear model techniques.