The paper contains a comparison of the results of calculation and experiment for the IOHNAP alloy steel. Specimens made of this steel were subjected to uniaxial constant-amplitude and random loading with both zero and non-zero mean values of loading. For determination of the steel fatigue life, the energy parameter including the mean value of loading was proposed. Under random loading, cycles were counted with the rain flow algorithm, and fatigue damage was accumulated with the Palmgren-Miner hypothesis. For the registered stress histories, elastic-plastic strains were calculated with the kinematic hardening model proposed by Mróz.

The aim of the paper is to present a procedure for generating service loading for fatigue tests of materials and structures. The generated loading characterizes desired functions of probability distribution and autocorrelation. The proposed numerical procedure uses MATLAB toolboxes and consists of three steps: (a) generation of a sequence of real numbers with the desired autocorrelation function and with any probability distribution function; (b) generation of loading history with the desired probability distribution function; (c) rearrangement of loading history (mentioned in item b) based on a sequence of real numbers with the desired correlation (mentioned in item a).

The paper contains a review of energy-based multiaxial fatigue failure criteria for cyclic and random loading. The criteria for cyclic loading have been divided into three groups, depending on the kind of strain energy density per cycle which is assumed as a damage parameter. They are: a) criteria based on elastic strain energy for high-cycle fatigue, b) criteria based on plastic strain energy for low-cycle fatigue. and c) criteria based on the sum of plastic and elastic strain energies for both low- and high-cycle fatigue. The criterion for random loading is based on the new definition of energy parameter which distinguishes plus and minus signs in history or specific work of stress on strain along chosen directions in the critical fracture plane. The criteria which rake into account strain energy density in the critical plane dominate in the energy description or multiaxial fatigue. Parameters dependent on loading and factors dependent on a kind or marcrial and inlluencing selection of the critical plane have been given. The author presented the mathematical models or the criteria and next distinguished those including influence of mean stresses and stress gradients as well as proportional and non-proportional loading. It has been emphasised that the generalized criterion of maximum shear and normal strain energy density in the critical plane seems to be the most efficient in practice and it should be developed and verified in a future.

The critical plane orientations determined with account for maximum value of energy density parameters and the weight function method were compared to experimental fatigue fracture plane orientations. Energy density parameters used in two multiaxial fatigue failure criteria, i.e. (i) criterion of the maximum normal strain energy density on the critical plane and (ii) criterion of the maximum shear strain energy density on the critical plane were employed. In the other method, the weight functions were formed on the basis of energy parameters. These two methods were verified by experimental tests of 1802A steel. The material was subjected to cyclic and random bending, torsion and combined bending with torsion with different coefficients of cross correlation between normal and shear stresses. The calculated results are satisfactory for both methods.