Bernard Dumon

Numerical pressure prediction algorithm of superplastic forming processes using 2D and 3D models

The paper describes a finite element model allowing for the prediction in a rapid way the optimum pressure cycle law, the deformed shapes, the distributions of the strain rate and the evolution of the thickness during superplastic forming processes. During the calculation, a pressure-cycle control algorithm is used in the analysis which keeps track of the maximum strain rate sensitivity index m, and adapts the pressure law applied in order to approach as nearly as possible the optimal pressure time history law. The purpose is not to follow the target exactly but obtain a practical pressure time history in a low computation cost time. To compare the performance of 2D and 3D approaches, two analyses have been performed using 2D and 3D finite element modeling. The numerical results are compared with the experimental ones to verify the validity of the pressure algorithm control.

Density and temperature dependence of spectral moments in depolarized light scattering by rare gases

We present an extensive study of the first three even moments M 2n of depolarized light scattering (DLS) spectra of rare gases argon and xenon. Most of our results were obtained by molecular dynamics and Monte Carlo computer simulation methods, which were used to calculate the moments for pairwise-additive intermolecular pair potential and interaction-induced polarizability models over a wide density range and at two temperatures, one above and one below the critical point. Simulation was used to determine the dependence of the moments on the pair polarizability anisotropy β(r) by comparing the results for the first order dipole-induced dipole (DID) model with those obtained for a semiempirical model of β(r), recently proposed by Meinander, Tabisz and Zoppi (MTZ), which includes electron overlap, dispersion and second order DID interaction contributions. Comparison with experiment indicates that the MTZ model for β(r) is clearly superior. Moments of increasing order are found to be increasingly more sensi...

Reliability estimation by Bayesian method: definition of prior distribution using dependability study

Abstract In this article, we present an evaluation method of the system reliability using Bayesian statistics and data feedback. In particular, we deal with the problem of the prior definition. A method is proposed to build the prior distribution from the dependability study (risk evaluation study).

Determining the shape parameter of a Weibull distribution from mechanical damage models

The aim of this paper is to show that it is possible to determine the shape factor of the Weibull distribution relating to a mechanical damage model. In particular, we define the shape parameter linked to damage by limited fatigue. Also, the parameters of distribution of cycles number to failure is characterized using the limited fatigue approach by the Basquin model. After, the Weibull parameters (/spl beta/ and /spl eta/) are estimated by the method of moments. This work is currently used for other modes in order subsequently to obtain more precise conclusions when one analyses a Weibull sheet.

Reliability analysis for complex industrial real-time systems: application on an antilock brake system

The reliability analysis of industrial systems is a very important engineering issue, in order to guarantee their functional behavior. Most of the critical failures are generated by the interactions between the sub-systems, implemented in different technologies, e.g. mechanics, electronics, and software. Therefore the analysis of the system as a whole is not enough and it is necessary to study all the interactions in order to estimate the system reliability.

Comparative study of accelerated testing models, applications in mechanics

Accelerated tests are becoming more and more popular in industry due to the need for obtaining life-data and to get more failures quickly. The article presents a comparative and critical analysis of some methods of accelerated testing (parametric, semi-parametric) in the case of mechanics. A simulation of a mechanical-damage model (fatigue) illustrates our analysis. The results obtained reveal the limitations and possible applications of these methods in mechanics.

Statistical analysis of accelerated experiments in mechanics using a mechanical accelerated life model

Accelerated testing of mechanical products offers great potential for development in reliability life testing. Unfortunately, difficulties met in accelerated life testing have limited its applications and acceptance. This paper presents a discussion of the different approaches to accelerate life of mechanical materials. The importance of failure mechanism models and the separate treatment of failure mechanisms during accelerated life tests are discussed. We formulate a new mechanical accelerated testing life model called Mecha-Statistical model. This model, considered as parametric, is constructed on the base of an appropriate mechanical damaging, model (fatigue). Like most traditional accelerated testing models, the Mecha-Statistical model consists of a life distribution that describes the lifetimes of components and relationships between the parameters of this distribution and stress. We suggest a method to construct the lognormal distribution based on the Basquin model that describes fatigue data and integrates mechanical characteristics of the product. The appropriate stress parameters and their limits are determined and studied for the failure mechanism under consideration (fatigue). They are selected so that the failure mode and mechanism generated under nominal stress are also generated under the accelerating stress. Assuming that number of cycles to failure has a lognormal distribution, the relationship between reliability and mechanical characteristics were modeled and used to estimate reliability. Optimum parameters of accelerated testing models were determined by optimization methods. These parameters were used to make comparison between our Mecha-Statistical model and the other parametric and semi parametric accelerated life models. The results obtained reveal that theses models do not fit the fatigue data better and that our model is useful and beneficial. The Mecha-Statistical model describes perfectly the lifetime testing data of fatigue tests because it depends on the product, the test method and the accelerating stress.

Reliability assessment of mechatronic systems: operating field data analysis

The reliability analysis of complex mechatronic systems is a very important engineering issue, in order to guarantee their functional behavior. We propose the evaluation of mechatronic systems reliability using censored data for operating field for different technologies, e.g. mechanics, electronics, and software. For ultra reliable system for which we have little data, we use an estimation method, stochastic expectation maximization (SEM), to increase the evaluation accuracy. The SEM method is used to estimate the reliability parameters with better accuracy than with maximization likelihood method.

Accelerated testing based on a mechanical-damage model

This article presents a method of accelerated testing for products under stress in order to estimate their level of reliability under normal conditions. To this end, the authors suggest an accelerated testing law testing based on a mechanical model of damage in order to be close to the failure mechanism. In particular, they propose a method based on a model of crack propagation under stress.

Accelerated test with unknown acceleration model

This paper discuss accelerated life testing (ALT) methods for life time distribution of highly reliable components. To define the reliability of components at a nominal stress level (s/sub 0/) when no acceleration model is known, a sample population often must be tested until many failures occur leading to extensive test time. In order to reduce the required time to obtain these failures, we suggest exposing the sample to a higher stress level (s/sub 1/) in the beginning of the test. Then, the surviving specimens are subjected to a second step at the nominal stress level (s/sub 0/). For a Weibull lifetime distribution model with unknown parameters, it has been shown that this step-stress test can lead to a great reduction in test time with the same accuracy for the estimation of parameters. This accuracy is compared to those obtained with existing methods of accelerated test processing.