For the past several decades there have been many attempts to improve suspension performance due to its importance within vehicle dynamics. The suspension system main functions are to connect the chassis to the ground, and to isolate the chassis from the ground. To improve upon these two functions, large amounts of effort are focused on two elements that form the building blocks of the suspension system, stiffness and damping. With the advent of new technologies, such as variable dampers, and powerful microprocessors and sensors, suspension performance can be enhanced beyond the traditional capabilities of a passive suspension system. Recently, Yin et al. [1, 2] have developed a novel dual chamber pneumatic spring that can provide tunable stiffness characteristics, which is rare compared to the sea of tunable dampers. The purpose of this thesis is to develop a controller to take advantage of the novel pneumatic spring’s functionality with a tunable damper to improve vehicle dynamic performance. Since the pneumatic spring is a slow-acting element (i.e. low bandwidth), the typical control logic for semi-active suspension systems are not practical for this framework. Most semi-active controllers assume the use of fast-acting (i.e. high bandwidth) variable dampers within the suspension design. In this case, a lookup table controller is used to manage the stiffness and damping properties for a wide range of operating conditions. To determine the optimum stiffness and damping properties, optimization is employed. Four objective functions are used to quantify vehicle performance; ride comfort, rattle space (i.e. suspension deflection), handling (i.e. tire deflection), and undamped sprung mass natural frequency. The goal is to minimize the first three objectives, while maximizing the latter to avoid motion sickness starting from 1Hz and downward. However, these goals cannot be attained simultaneously, necessitating compromises between them. Using the optimization strength of genetic algorithms, a Pareto optima set can be generated to determine the compromises between objective functions that have been normalized. Using a trade-off study, the stiffness and damping properties can be selected from the Pareto optima set for suitability within an operating condition of the control logic. When implementing the lookup table controller, a practical method is employed to recognize the road profile as there is no direct method to determine road profile. To determine the road profile for the lookup table controller, the unsprung mass RMS acceleration and suspension state are utilized. To alleviate the inherent flip-flopping drawback of lookup table controllers, a temporal deadband is employed to eliminate the flip-flopping of the lookup table controller. Results from the semi-active suspension with tunable stiffness and damping show that vehicle performance, depending on road roughness and vehicle speed, can improve up to 18% over passive suspension systems. Since the controller does not constantly adjust the damping properties, cost and reliability may increase over traditional semi-active suspension systems. The flip-flopping drawback of lookup table controllers has been reduced through the use of a temporal deadband, however further enhancement is required to eliminate flip-flopping within the control logic. Looking forward, the novel semi-active suspension has great potential to improve vehicle dynamic performance especially for heavy vehicles that have large sprung mass variation, but to increase robustness the following should be considered: better road profile recognition, the elimination of flip-flopping between suspension states, and using state equations model of the pneumatic spring within the vehicle model for optimization and evaluation.

For the past several decades there have been many attempts to improve suspension performance due to its importance within vehicle dynamics. The suspension system main functions are to connect the chassis to the ground, and to isolate the chassis from the ground. To improve upon these two functions, large amounts of effort are focused on two elements that form the building blocks of the suspension system, stiffness and damping. With the advent of new technologies, such as variable dampers, and powerful microprocessors and sensors, suspension performance can be enhanced beyond the traditional capabilities of a passive suspension system. Recently, Yin et al. [1, 2] have developed a novel dual chamber pneumatic spring that can provide tunable stiffness characteristics, which is rare compared to the sea of tunable dampers. The purpose of this thesis is to develop a controller to take advantage of the novel pneumatic spring’s functionality with a tunable damper to improve vehicle dynamic performance. Since the pneumatic spring is a slow-acting element (i.e. low bandwidth), the typical control logic for semi-active suspension systems are not practical for this framework. Most semi-active controllers assume the use of fast-acting (i.e. high bandwidth) variable dampers within the suspension design. In this case, a lookup table controller is used to manage the stiffness and damping properties for a wide range of operating conditions. To determine the optimum stiffness and damping properties, optimization is employed. Four objective functions are used to quantify vehicle performance; ride comfort, rattle space (i.e. suspension deflection), handling (i.e. tire deflection), and undamped sprung mass natural frequency. The goal is to minimize the first three objectives, while maximizing the latter to avoid motion sickness starting from 1Hz and downward. However, these goals cannot be attained simultaneously, necessitating compromises between them. Using the optimization strength of genetic algorithms, a Pareto optima set can be generated to determine the compromises between objective functions that have been normalized. Using a trade-off study, the stiffness and damping properties can be selected from the Pareto optima set for suitability within an operating condition of the control logic. When implementing the lookup table controller, a practical method is employed to recognize the road profile as there is no direct method to determine road profile. To determine the road profile for the lookup table controller, the unsprung mass RMS acceleration and suspension state are utilized. To alleviate the inherent flip-flopping drawback of lookup table controllers, a temporal deadband is employed to eliminate the flip-flopping of the lookup table controller. Results from the semi-active suspension with tunable stiffness and damping show that vehicle performance, depending on road roughness and vehicle speed, can improve up to 18% over passive suspension systems. Since the controller does not constantly adjust the damping properties, cost and reliability may increase over traditional semi-active suspension systems. The flip-flopping drawback of lookup table controllers has been reduced through the use of a temporal deadband, however further enhancement is required to eliminate flip-flopping within the control logic. Looking forward, the novel semi-active suspension has great potential to improve vehicle dynamic performance especially for heavy vehicles that have large sprung mass variation, but to increase robustness the following should be considered: better road profile recognition, the elimination of flip-flopping between suspension states, and using state equations model of the pneumatic spring within the vehicle model for optimization and evaluation.

The purpose of this study is to mitigate the compromise between ride comfort and handling of a small single seat off-road vehicle known as a Baja. This has been achieved by semi-active control of the suspension system containing controllable magnetorheological (MR) dampers and passive hydro-pneumatic spring-damper units. MR fluid is a viscous fluid whose rheological properties depend on the strength of the magnetic field surrounding the fluid, and typically consists of iron particles suspended in silicone oil. When a magnetic field is applied to the fluid, the iron particles become aligned and change the effective viscosity of the fluid. The use of MR fluid in dampers provides variable damping that can be changed quickly by controlling the intensity of the magnetic field around the fluid. Various benefits associated with the use of MR dampers have led to their widespread implementation in automotive engineering. Many studies on conventional vehicles in the existing literature have demonstrated the conflicting suspension requirements for favourable ride comfort and handling. Generally, soft springs with low damping are ideal for improved ride comfort, while stiff springs with high damping are required for enhanced handling. This has resulted in the development of passive suspension systems that provide either an enhanced ride quality or good drivability, often targeting one at the expense of the other. The test vehicle used for this study is distinct in many ways with multiple characteristics that are not commonly observed in the existing literature. For instance, the absence of a differential in the test vehicle driveline causes drivability issues that are aggravated by increased damping. The majority of existing MR damper models in the literature are developed for uniform excitation and re-characterisation of model parameters is required for changes in input conditions. Although recursive models are more accurate and applicable to a wider range of input conditions, these models require measured force feedback which may not always be available due to limitations such as packaging constraints. These constraints required the development of alternative MR damper models that can be used to prescribe the current input to the damper. In this study parametric, nonparametric and recursive MR damper models have been developed and evaluated in terms of accuracy, invertibility and applicability to random excitation. The MR damper is used in parallel with passive damping as a certain amount of passive damping is always present in suspension systems due to friction and elastomeric parts. Most of the existing studies on suspension systems have been performed using linear two degree of freedom vehicle models that are constrained to specific conditions. Usually these models are implemented without an indication of the ability of these models to accurately represent the vehicles that these studies are intended for. For this study, a nonlinear, three-dimensional, 12 degrees of freedom vehicle model has been developed to represent the test vehicle. This model is validated against experimental results for ride comfort and handling. The MR damper models are combined with the model of the test vehicle, and used in ride comfort and handling simulations at various levels of passive damping and control gains in order to assess the potential impact of suspension control on the ride quality and drivability of the test vehicle. Simulation results show that lower passive damping levels can significantly improve the ride comfort as well as the handling characteristics of the test vehicle. Furthermore, it is observed that additional improvements that may be obtained by the implementation of continuous damping control may not be justifiable due to the associated cost and complexity. / Dissertation (MEng)--University of Pretoria, 2013. / Mechanical and Aeronautical Engineering / unrestricted

The diploma thesis is focused on the development of the concept of a mechanical device that would meet the necessary requirements to achieve a successful simulation of reduced gravity or microgravity on the tested body at its impact. The choice of a suitable mechanism is based on a literary research of all available variants and highlight of their specifics. Target values are created for the selected method of testing, upon reaching which the concept will be considered successful. In order to verify the functionality of the concept, a mechanism is assembled on a smaller scale. The individual components of the mechanism are tested and carefully selected. All essential physical processes taking place in the system are mathematically described and combined into a MATLAB script. A separate application is created to calculate the behavior of the system based on the specified input parameters.