ANALYSIS OF FAILURE FLOWS OF MOTOR-VEHICLE COMPONENTS

Abstract

The article examines the forecasting of reliability indicators for motor-vehicle components, assemblies, and systems during design and operation. The calculation- based forecasting framework is shown to require a hierarchical representation of the product, load spectra, physical damage models, failure and limit-state criteria, and procedures for assessing and correcting the forecast. Because reliability indicators are stochastic, an exact estimate of mean service life is often impossible, particularly when information on new designs, materials, and operating modes is incomplete. General and special forecasting methods are classified, and a combined approach integrating extrapolation, statistical, and structural-functional models is substantiated. Directed state graphs are considered for an individual vehicle and a fleet, describing operation, standby, diagnosis, maintenance, and restoration. Wear is represented as a monotonically increasing nonstationary random process whose mean and variance vary with time. For common wear approximations, relationships are obtained for the probability density of service life. The case in which the limiting wear is a random variable is investigated by Monte Carlo simulation, linearization, and numerical integration. A numerical example demonstrates close agreement between the moments of the service-life distribution obtained by the three methods. The results can support the selection of design alternatives and maintenance and repair strategies, refinement of operating parameters and modes, and planning of spare-parts demand. The proposed framework links reliability prediction with decision-making under uncertainty and therefore provides a methodological basis for managing vehicle reliability throughout the life cycle. It also improves the consistency of decisions made under risk and incomplete information.