Search

The highly nonlinear characteristics of dry friction damped suspension components used in railway freight vehicles increases the modelling challenges in Multibody Simulations (MBS) software packages such as VAMPIRE, SIMPACK and ADAMS VI-RAIL. These system nonlinearities then lead to lower confidence levels in modelling methodologies and therefore, more uncertainty in the simulation results produced by these models. The main reasons for the low confidence levels associated with friction modelling methodologies in the context of railway freight vehicles, are due to the lack of reliable test data to validate the friction models used in the primary/secondary suspension. Therefore, there is an industry requirement to better understand the in service/maintenance requirements of these components, first from a testing perspective and then from a modelling one. Once this current gap in knowledge and understanding is bridged, the results generated from the simulation models could then be used in the future to replace on-track/laboratory testing, which would bring significant cost benefits to the rail freight industry. The work presented in this thesis reviews the current friction testing and modelling methodologies as well as the simplifying assumptions that are implemented in commercial MBS software packages. The most common friction model that is used in railway vehicle MBS for example, is the Coulomb model of dry friction. This model however, is not capable of describing stick-slip behaviour, pre-sliding displacements or the viscoelastic effects of friction surfaces when they start to yield under extreme loading conditions. The Coulomb friction model also has a discontinuity in the vicinity of zero velocity, which can cause numerical issues in MBS as the friction force in this region, is non-differentiable. To handle the discontinuity, a variety of regularisation methods are generally implemented in the Coulomb model which are further discussed in the literature review. The impact of these regularisation methods on the validity of MBS results however are currently not fully understood or justified, as these tend to be mathematical smoothing functions that are not actually related to the mechanical properties of the friction surfaces. Therefore, to better understand the limitations of the Coulomb friction model and its application, as well as the different regularisation methods that are implemented in freight vehicle secondary suspension friction models, several scale model benchmark tests have been carried out on a variety of rail industry standard secondary suspension wear liner components. The critical modelling parameters that are required to develop a comprehensive/optimised freight vehicle secondary suspension friction model were then calculated from the benchmark test data. These parameters include the torsional stiffness and damping properties of the centre plate/wear liner components, as well as the Coefficient of Friction (COF) values and frequency signatures of the secondary suspension system. From this research, three novel secondary suspension friction models are proposed that have been developed to work in conjunction with the SIMPACK software package via the co-simulation functionality. These models include the novel local/global friction models, which are used together in MBS to describe the hysteresis and frequency responses of the centre plate/wear liner components as well as increases in these parameters as a function of time and the number of running cycles.

Custom algorithms and stand-alone software have been originated that allow the automatic generation of complex Magnetic Equivalent Circuit (MEC) models from the description of a rotary machines physical geometry and materials. The models are automatically generated in a format that can be utilised in a standard circuit simulator package, such as SPICE, allowing the simulation of the full electrical, magnetic and mechanical interaction. Fully interactive machine models have been developed by extending Magnetic Equivalent Circuits (MECs) to include mechanical torque by the application of the virtual work principle. Using a common finite element B-H formula, the models are further enhanced by the incorporation of non-linear permeance. These equivalent models use standard circuit components that are available in most circuit solver packages making them transportable and vendor independent. A novel stand-alone software package has been created that simplifies the generation of the complex MEC models. By describing the machine geometry, flux paths and material characteristics in a simple file an electrical equivalent circuit can be automatically generated at different levels of abstraction, depending on the input information. The software allows the machine's geometry to be viewed and, after simulation, the flux information can be back annotated into the geometry viewer for further design refinement. The modelling technique has been verified by automatically generating and simulating a MEC model of a 3-phase induction motor; this was checked against practical results obtained from a test-bed consisting of an induction motor and associated load. A further validation of the MEC modelling technique checks for consistency by ascertaining that the energy flow between the electrical, magnetic and mechanical sections balance for a synchronous generator and synchronous motor. The automatic production of complex Magnetic Equivalent Circuits of rotary machines produces a realistic model giving a high degree of confidence in the simulation results when combined with the power and control electronics in a SPICE type simulator.