The shape of a rainflow matrix is complex and cannot be approximated by a simple distribution function. In this paper, the Weibull-normal mixture distribution is used, for which the number of components and unknown parameters are required to be estimated. The scope of the paper is to estimate the number of components and unknown parameters using the FlexMix and REBMIX algorithms, and compare their results. The results are then used in Goodman and Walker mean stress correction methods. This correction is not made as a point-to-point transformation, where the information about the distribution function of the rainflow matrix is lost. Instead, the used distribution function of the rainflow matrix with estimated parameters is transformed in accordance with Goodman and Walker mean stress correction methods. With this procedure, the probability density of the equivalent stress amplitude is immediately obtained, and the information about the distribution function of the rainflow matrix is not lost.
COBISS.SI-ID: 13747227
A method is presented that enables predictions of long-term creep-rupture strength based on a small sample of short-term experimental results. Smoothed bootstrapping in combination with the most commonly used time-temperature parameters is used for the determination of the optimal values of coefficients. For ten metals, creep%rupture strengths are evaluated for both full-size and sub-size data sets. Predictions in the case of small data sets always lie on the conservative side, whereas the confidence interval of predicted strengths decreases with an increasing number of experimental results. However it is shown here that for the evaluated materials, usable interim predictions of creep-rupture strengths can be achieved.
COBISS.SI-ID: 13803803
This article describes how a selected materialʼs fatigue-life-curve model influences the calculated reliability of a structure subjected to a dynamic loading. A uni-axially loaded structural beam with a fully-reversal constant loading amplitude was considered. The reliability for a certain number of cycles-to-failure was calculated as a cross-section of the probability distributions representing the load-amplitude scatter and the scatter of the materialʼs fatigue-life curve. The probability density function (PDF) of the loading amplitude was modelled by a uniform and a Gaussian PDF. The scattered fatigue-life curve was modelled by a conditional two-parametric Weibullʼs PDF. Its parameters were estimated using two procedures: (i) a two-phase procedure and (ii) a direct procedure. Following the two-phase procedure a conditional PDF of the number of cycles-to-failure was estimated first and then converted into a corresponding conditional PDF of the stress amplitudes. In the direct procedure the conditional PDF of the stress amplitudes was modelled directly from the fatigue-life data. The two procedures were tested on 12 sets of simulated fatigue-life data and a set of experimental fatigue-life data. The two fatigue-life-curve models for the experimental data set were applied for calculating the reliability for the selected structural beam.
COBISS.SI-ID: 14035739
The existing modified Locati step-test was improved by pairing it with two evolutionary algorithms that were used to minimize the newly defined cost function that measures the deviation of the "best-fit point" trace from the selected target number of load cycles. The main advantage of the new procedure over the conventional Locati step-test is that the slopes and intercepts of the materialʼs or the componentʼs S-N curves do not need to be known in advance in order to estimate the fatigue strength at a selected number of load cycles. The presented procedure was tested on two cases with a different target number of load cycles. The results indicate that the presented procedure is adequate to the conventional Locati step-test procedure.
COBISS.SI-ID: 13999387
Experimental and numerical analysis of a compression test carried out on samples of as received and thermally treated beech (Fagus sylvatica L.)wood is presented. In a normal climate, samples with the dimensions of 20 x 20 x 20 mm were exposed to static compressive loads parallel and transverse to the grain. Afterwards, the test was mode lled using the finite element method. It was confirmed that, after thermal modification, the wood’s density decreased and the stiffness in both tested directions increased. After the thermal treatment, the strength of beech wood increased in the direction parallel to the grain and decreased in the direction transverse to the grain. Based on the comparison of experimental and numerical results, it is possible to use the hyperelastic constitutive law to reasonably model the force and displacement obtained in the compression test samples.
COBISS.SI-ID: 2468745