Dieter Bestle

An Orthogonal Complement Matrix Formulation for Constrained Multibody Systems

A method is proposed for the automatic generation of an orthogonal complement matrix to the constraint matrix for the dynamic analysis of constrained multibody systems. The clue for this method lies in the determination of local constraint matrices and their orthogonal complements relative to the local reference frames of particular constrained points. These matrices are then transformed into the systems configuration space in order to form the final constraint matrix and its orthogonal complement. The avoidance of singularities in the formulation is discussed. The method is specially suited for the dynamic analysis of multibody systems with many constraints and/or closed-loops.

Optimal shift control for automatic transmission

In general, gearshift is related to change of acceleration due to the changing gear ratio. Modern double-clutch transmissions allow for shaping the acceleration transition by controlling the torques transmitted by the clutches. Thus, the question arises about an optimal transition law for the acceleration. The paper demonstrates that jerk and change of jerk may be considered as major sources of discomfort. Thus, a bi-criterion optimization problem is formulated for finding an optimal acceleration transition and an associated shift control. The problem is solved by analytical and polynomial approaches, and a theoretically optimal solution is shown. The latter is applied to a simple simulation model of a double-clutch transmission to demonstrate its applicability.

OPTIMIZATION OF DAMPING CHARACTERISTICS IN VEHICLE DYNAMICS

In vehicle dynamics, designing the characteristics of shock absorbers is an important subproblem which determines the overall behavior of the vehicle to a high extent. Their design can be supported by numerical optimization. In the paper the problem of an adequate description is discussed where the mathematical function type used is determined by physical effects. In order to apply efficient optimization algorithms, sensitivities with respect to state variables and control nodes of the characteristics are computed. A suspension design problem based on a multibody system model of a vehicle is formulated and solved by multicriterion optimization. Results for different preferences on comfort and riding safety are shown.

Modern Compressor Aerodynamic Blading Process Using Multi-Objective Optimization

Aerodynamic design of an axial compressor is a challenging design task requiring a compromise between contradicting requirements like wide operating range, high efficiency, low number of stages and high surge margin. Therefore, the design process is typically subdivided into a sequence of subproblems where the blading design is a key process. According to flow conditions, which result from throughflow calculations on axis-symmetric stream surfaces, 2-dimensional blade profiles have to be designed, which then may be stacked along a radial stacking line in order to find the 3D-blade geometry. The design of the blade sections is rather time consuming due to many iterations with different programs. Usually a geometry generation tool is used to describe the blade sections which are then evaluated by a blade-to-blade CFD solver. The quality of a single blade section is typically characterized by the overall loss at design flow conditions and the working range determined by an amount of loss increase due to incidence variation. The aerodynamic performance of the final airfoils and thus of the whole compressor depends significantly on the design of the individual blade sections. In this investigation an automated multi-objective optimization strategy is developed to find best blade section geometries with respect to loss and working range. The multi-objective optimization approach provides Pareto-optimal compromise solutions at reasonable computational costs outperforming a given Rolls-Royce datum design which has been ‘optimized’ manually by a human design engineer.Copyright

Optimisation of lateral car dynamics taking into account parameter uncertainties

Simulation studies on an active all-wheel-steering car show that disturbance of vehicle parameters have high influence on lateral car dynamics. This motivates the need of robust design against such parameter uncertainties. A specific parametrisation is established combining deterministic, velocity-dependent steering control parameters with partly uncertain, velocity-independent vehicle parameters for simultaneous use in a numerical optimisation process. Model-based objectives are formulated and summarised in a multi-objective optimisation problem where especially the lateral steady-state behaviour is improved by an adaption strategy based on measurable uncertainties. The normally distributed uncertainties are generated by optimal Latin hypercube sampling and a response surface based strategy helps to cut down time consuming model evaluations which offers the possibility to use a genetic optimisation algorithm. Optimisation results are discussed in different criterion spaces and the achieved improvements confirm the validity of the proposed procedure.

3D CFD Compressor Map Computation Process Accounting for Geometry Changes due to Off-Design Operation Loads

Compressor maps of aero engines show the relation between corrected inlet mass flow and total pressure ratio for various engine speeds. Different speed lines represent different operating conditions of the compressor, where especially operating bounds like surge and choke are important for the design process. Typically, 3D CFD compressor maps are computed with the so called hot geometry given for the aerodynamic design point. However, in reality airfoil shapes will change for different engine speeds and gas loads resulting in twisted airfoils and changed tip clearances. Thus, using the nominal hot geometry for the whole compressor map is not fully correct. In order to obtain higher quality performance maps these effects need to be considered. The paper shows a process for computing compressor maps with 3D CFD, where strucural deformations of the blade due to varying speeds and gas loads are taken into account by blade morphing. This process is applied to a 1.5-stage compressor showcase.Copyright

Using gestures to interactively modify turbine blades in a Virtual Environment

Working in highly immersive Virtual Environments (VEs) demands special interaction techniques. If the user needs to interact three-dimensionally, traditional input devices such as keyboard or mouse are not appropriate. Gesture-based interaction is an alternative to expensive and cumbersome input hardware. In this paper a device-less, gesture-based interaction is proposed which enables to interactively modify turbine blades with predefined handles and observe the impact on the fluid flow in a VE. By using a GPU-accelerated computational fluid dynamics (CFD) solver the fluid flow for the new blade geometry is computed within a few seconds. Such a rapid analysis process shall enable turbine engineers to optimize their designs interactively.

Application of Optimization Methods to Controller Design for Active Suspensions

Abstract By offering good ride safety and ride comfort to passenger cars, active suspensions have attracted more and more attention of investigators. Numerous approaches of designing controllers for active suspension systems have been introduced mostly restricted to linear time-invariant systems. In this paper, an approach is proposed where an extended linear-quadratic regulator (LQR) control for linear systems with measurable disturbances is combined with a multi-criterion optimization (MCO) procedure. This allows to directly taking into account criteria with physical interpretation, to guide decision-making in case of contradicting criteria, and to reduce the number of design variables of the MCO problem significantly. The approach is efficient enough to extend it to gain-scheduling control of a linear-parameter-varying spatial car model resulting from changing speed and track curvature. The effectiveness of the designed controller with respect to ride safety and ride comfort is demonstrated through the simulation of a double-lane-change maneuver, where the path itself is found by an optimization procedure.

A Non-Dimensional Quasi-3D Blade Design Approach With Respect to Aerodynamic Criteria

The competition between aero-engine manufacturers has increased dramatically in the last decades. Saving computational time within the design process, which is equivalent to saving money, is of major importance for the industry. Talking about the aerodynamic compressor blading process, it becomes indispensable to go for new or alternative ways in designing blades in order to fulfill raised performance demands. The focus of this paper, therefore, is to propose a quasi-3D aerodynamic design concept with extended and improved parameterization of the aerofoil in order to support the industrial blading process. A Bezier-surface is selected to parameterize the non-dimensional camber-line angle distribution along the blade chord from leading to trailing edge over the entire blade height in radial direction. Starting from scratch, the geometric blade build-up is completed by superposing the resulting camber-line with a given thickness distribution. For additional increase of design freedom, Bezier-curves are used to radially parameterize blade inlet and outlet angles in their dimensionless form. The chosen parameterization of these distributions guarantees smooth blade shapes and geometry distributions with a minimum of design parameters. For optimization purpose it is essential to get performance information on the entire blade, however, with minimal computational effort. Facing this challenge, aerodynamic blade performance is evaluated by a two-dimensional blade-to-blade flow solver for specific sections on different radial blade heights. In order to speed up the blade design process, the flow calculations are realized by a distributed computing concept on a Linux high-performance cluster. All investigations are carried out for highly loaded controlled diffusion blades which are taken from an existing industrial research application. Since selected criteria such as mean loss at design point conditions and working range for off-design flow conditions represent contradicting design goals, the blade design problem is solved by means of a multi-objective problem formulation and a stochastic optimization algorithm. As a result Pareto-optimal trade-off solutions between conflicting design goals are shown where the design engineer can choose from according to his specific preferences.Copyright

Optimization of Air Distribution in a Preliminary Design Stage of an Aero-Engine Combustor

The design of a modern aero-engine combustor is strongly driven by severe emission regulations. The paper describes an automated preliminary aero-thermal design process for a rich-burn combustor by combining different low fidelity analysis tools in order to speed up the preliminary design loop and provide improved combustor designs. Design evaluation is performed by a knowledge-based preliminary design tool coupled with a network solver. The preliminary design tool provides a 2D geometry model and cooling layout based on industrial in-house design rules. The 1D network solver then calculates the air distribution inside the combustor for two state-of-the-art combustor cooling schemes, i.e., single skin cooling and double skin tiled cooling. The computed air distribution is subsequently used to predict emissions which are minimized by using a genetic multi-objective optimization algorithm. As a result, better designs are obtained with the prescribed approach compared to a human reference design.