Development of an algorithm for piston secondary dynamics and skirt-liner contact and its use in multi-objective piston design under uncertainty

Low vibration and noise level in internal combustion engines has become an essential part of the design process. It is well known that the piston assembly can be a major source of engine mechanical friction and cold start noise if not designed properly. The piston secondary motion and piston-bore contact pattern are critical in piston design because they affect the skirt-to-bore impact force and, therefore, how the piston impact excitation energy is damped, transmitted, and eventually radiated from the engine structure as noise. An analytical method is proposed in this research for simulating piston secondary dynamics and piston-bore contact for an asymmetric half piston model. The method includes several important physical attributes such as bore distortion effects due to mechanical and thermal deformation, inertia loading, piston barrelity and ovality, piston flexibility, and skirt-to-bore clearance. The method accounts for piston kinematics, rigid-body dynamics, and skirt flexibility. A time integration algorithm is used based on a modified Newmark-Beta method for nonlinear systems. A contact algorithm using a complementarity method determines the contact characteristics between the flexible piston and a rigid liner during the engine cycle. The most important aspects of the resultant model are efficiency, accuracy, and robustness. The developed analytical model is also used in piston design. A parametric study is conducted first, considering engine operating parameters and a large number of piston design variables in order to understand the most important piston design aspects. Subsequently, a detailed sensitivity analysis is performed using principles from the design of experiments method, to find the key design variables affecting the piston friction, noise, and scuffing performance measures. The used sample points in the design of experiments approach are also used to create approximation models (metamodels) between the most dominant input design variables and the three piston performance measures. The metamodels are used in deterministic and reliability-based design optimization to optimize the piston design considering the trade-offs among the multiple objectives of friction, noise, and scuffing in the presence, or not, of uncertainty. The multi-objective optimization process proposed in this research allows the designer to choose an optimal design, among the many on the Pareto frontier of optimal solutions, which satisfies his/her subjective preferences. Simple examples and a detailed piston design example, demonstrate all concepts and the advantages of the developed piston analytical model in design.