An efficient framework for the reliability-based design optimization of large-scale uncertain and stochastic linear systems

This paper is focused on the development of an efficient reliability-based design optimization algorithm for solving problems posed on uncertain linear dynamic systems characterized by large design variable vectors and driven by non-stationary stochastic excitation. The interest in such problems lies in the desire to define a new generation of tools that can efficiently solve practical problems, such as the design of high-rise buildings in seismic zones, characterized by numerous free parameters in a rigorously probabilistic setting. To this end a novel decoupling approach is developed based on defining and solving a limited sequence of deterministic optimization sub-problems. In particular, each sub-problem is formulated from information pertaining to a single simulation carried out exclusively in the current design point. This characteristic drastically limits the number of simulations necessary to find a solution to the original problem while making the proposed approach practically insensitive to the size of the design variable vector. To demonstrate the efficiency and strong convergence properties of the proposed approach, the structural system of a high-rise building defined by over three hundred free parameters is optimized under non-stationary stochastic earthquake excitation.     

A two-step method for analysis of linear systems with uncertain parameters driven by Gaussian noise

A two-step method is proposed to find state properties for linear dynamic systems driven by Gaussian noise with uncertain parameters modeled as a random vector with known probability distribution. First, equations of linear random vibration are used to find the probability law of the state of a system with uncertain parameters conditional on this vector. Second, stochastic reduced order models (SROMs) are employed to calculate properties of the unconditional system state. Bayesian methods are applied to extend the proposed approach to the case when the probability law of the random vector is not available. Various examples are provided to demonstrate the usefulness of the method, including the random vibration response of a spacecraft with uncertain damping model.     

On the efficacy of stochastic collocation,stochastic Galerkin,and stochastic reduced order models for solving stochastic problems

The stochastic collocation (SC) and stochastic Galerkin (SG) methods are two well-established and successful approaches for solving general stochastic problems. A recently developed method based on stochastic reduced order models (SROMs) can also be used. Herein we provide a comparison of the three methods for some numerical examples; our evaluation only holds for the examples considered in the paper. The purpose of the comparisons is not to criticize the SC or SG methods, which have proven very useful for a broad range of applications, nor is it to provide overall ratings of these methods as compared to the SROM method. Rather, our objectives are to present the SROM method as an alternative approach to solving stochastic problems and provide information on the computational effort required by the implementation of each method, while simultaneously assessing their performance for a collection of specific problems.     

A warranty forecasting model based on piecewise statistical distributions and stochastic simulation

This paper presents a warranty forecasting method based on stochastic simulation of expected product warranty returns. This methodology is presented in the context of a high-volume product industry and has a specific application to automotive electronics. The warranty prediction model is based on a piecewise application of Weibull and exponential distributions, having three parameters, which are the characteristic life and shape parameter of the Weibull distribution and the time coordinate of the junction point of the two distributions. This time coordinate is the point at which the reliability ‘bathtub’ curve exhibits a transition between early life and constant hazard rate behavior. The values of the parameters are obtained from the optimum fitting of data on past warranty claims for similar products. Based on the analysis of past warranty returns it is established that even though the warranty distribution parameters vary visibly between product lines they stay approximately consistent within the same product family, which increases the overall accuracy of the simulation-based warranty forecasting technique. The method is demonstrated using a case study of automotive electronics warranty returns. The approach developed and demonstrated in this paper represents a balance between correctly modeling the failure rate trend changes and practicality for use by reliability analysis organizations.     

Analysis of one-dimensional stochastic finite elements using neural networks

This paper explores the applicability of neural networks for analyzing the uncertainty spread of structural responses under the presence of one-dimensional random fields. Specifically, the neural network is intended to be a partial surrogate of the structural model needed in a Monte Carlo simulation, due to its associative memory properties. The network is trained with some pairs of input and output data obtained by some Monte Carlo simulations and then used in substitution of the finite element solver. In order to minimize the size of the networks, and hence the number of training pairs, the Karhunen–Loéve decomposition is applied as an optimal feature extraction tool. The Monte Carlo samples for training and validation are also generated using this decomposition. The Nyström technique is employed for the numerical solution of the Fredholm integral equation. The radial basis function (RBF) network was selected as the neural device for learning the input/output relationship due to its high accuracy and fast training speed. The analysis shows that this approach constitutes a promising method for stochastic finite element analysis inasmuch as the error with respect to the Monte Carlo simulation is negligible.     

Stochastic quasi-gradient based optimization algorithms for dynamic reliability applications

On one hand, PSA results are increasingly used in decision making, system management and optimization of system design. On the other hand, when severe accidental transients are considered, dynamic reliability appears appropriate to account for the complex interaction between the transitions between hardware configurations, the operator behavior and the dynamic evolution of the system. This paper presents an exploratory work in which the estimation of the system unreliability in a dynamic context is coupled with an optimization algorithm to determine the “best” safety policy. Because some reliability parameters are likely to be distributed, the cost function to be minimized turns out to be a random variable. Stochastic programming techniques are therefore envisioned to determine an optimal strategy. Monte Carlo simulation is used at all stages of the computations, from the estimation of the system unreliability to that of the stochastic quasi-gradient. The optimization algorithm is illustrated on a HNO₃ supply system.     

Using computer techniques for vibration damage estimation under stochastic loading using the Monte Carlo Method for aerospace applications

The main focus of this article is a review of legacy methods for vibration damage estimation under stochastic loading and extending research made by Dirlik and Bendat using two combined methods: FEM and Monte Carlo simulation, for which we used Python programming for aerospace applications. For some aircraft, regulated by the RTCA international aviation standard DO-160G (Environmental Conditions and Test Procedures for Airborne Equipment), stochastic loading is defined as one of the requirements. This article will focus on the stochastic loading impact on the fatigue life assessment made on a dummy sample, and frequency and time domain damage estimation shall be considered in parallel to compare both results. Additionally, dummy PSD responses shall be defined in the frequency domain for signal statistical parameters research. The article introduces Rainflow Cycle Counting methods in the frequency domain for procedures used commercially in aerospace applications. The first method introduced and developed further is the Dirlik method of Rainflow Cycle Counting in the frequency domain, which is the most popular method in commercial use. The second technique introduced and developed further was established by Bendat — the Narrow Band Method. The new empirical equation presented in this paper is the modification of the Narrow Band Method fitted for general use (narrow band, wide band, and white noise signals). A new approach for the integration of spectral moments is introduced in this paper, allowing for an accurate evaluation of the signal statistic parameters in the frequency domain for use in the modified Dirlik and Narrow Band methods. Research results also revealed new phenomena not researched by Dirlik, such as high vibration damage variation from stochastic loading, which depends on the frequency resolution (the block size used in Inverse Fourier Transformation). This discovery will be the subject of further study. Research results presented in this paper will also be utilised to combine stochastic and deterministic loading scenarios for military helicopters, as well as fighter aircraft, and will be the subject of further research.     

A stochastic multimode model for time–cost tradeoffs under management flexibility

In this paper, we present a useful bicriteria model for analyzing the time–cost tradeoffs that can be achieved when undertaking a project consisting of a single task. Some related mathematical results are also presented. The model assumes that several different processes may be used to undertake the task, defining different operating modes, and also assuming that managers may change the process in use after the task is started, depending on the way the task is developing. The model is easy to use, and it can be applied to a large number of real-life situations when the objectives are to minimize cost and time. We present some ideas for an algorithm aimed at identifying sets of relevant strategies from which the decision-maker may choose a suitable strategy. We also present several mathematical properties that may be useful for such an algorithm.     

Moment operations on random variables, with applications for probabilistic analysis

A general method is developed for conducting simple operations on random variables, avoiding difficult integrals and singularities, which must be overcome when obtaining exact solutions. For sum, difference and product operations, and combinations thereof, exact moments are first determined from the moments of the constituent variables. The method of orthogonal expansion, developed in the previous paper [Probabilistic Engineering Mechanics 2000;15:371–379], is then used to produce approximate probability density functions (PDFs). The quotient operation is also considered; it requires knowledge of the negative moments of the denominator variable. The quotient and difference operations are used in a first example to establish PDFs for the hazard quotient and excess wind loading on a concrete chimney. A second example demonstrates how the proposed method may be used as an alternative to Monte Carlo simulation for simple probabilistic risk calculations; a PDF for predicted contaminant concentration at a groundwater well compares favorably with a histogram obtained by simulation.     

A Monte Carlo simulation model for stationary non-Gaussian processes

A class of stationary non-Gaussian processes, referred to as the class of mixtures of translation processes, is defined by their finite dimensional distributions consisting of mixtures of finite dimensional distributions of translation processes. The class of mixtures of translation processes includes translation processes and is useful for both Monte Carlo simulation and analytical studies. As for translation processes, the mixture of translation processes can have a wide range of marginal distributions and correlation functions. Moreover, these processes can match a broader range of second order correlation functions than translation processes. The paper also develops an algorithm for generating samples of any non-Gaussian process in the class of mixtures of translation processes. The algorithm is based on the sampling representation theorem for stochastic processes and properties of the conditional distributions. Examples are presented to illustrate the proposed Monte Carlo algorithm and compare features of translation processes and mixture of translation processes.     

Parallel processing in computational stochastic dynamics

Studying large complex problems that often arise in computational stochastic dynamics (CSD) demands significant computer power and data storage. Parallel processing can help meet these requirements by exploiting the computational and storage capabilities of multiprocessing computational environments. The challenge is to develop parallel algorithms and computational strategies that can take full advantage of parallel machines. This paper reviews some of the characteristics of parallel computing and the techniques used to parallelize computational algorithms in CSD. The characteristics of parallel processor environments are discussed, including parallelization through the use of message passing and parallelizing compilers. Several applications of parallel processing in CSD are then developed: solutions of the Fokker–Planck equation, Monte Carlo simulation of dynamical systems, and random eigenvector problems. In these examples, parallel processing is seen to be a promising approach through which to resolve some of the computational issues pertinent to CSD.     

A study of stochastic fatigue crack growth modeling through experimental data

To capture the statistical nature of fatigue crack growth, many stochastic models have been proposed in the literature. These models may have been verified by only one data set, and therefore not appreciated by other fellow researchers. Part of the reason is the difficulty and time-consuming in obtaining the statistically meaningful fatigue crack growth data. In the present study, experimental work is carried out to obtain the fatigue crack growth data of a batch of 2024-T351 aluminum alloy specimens. A rather universal stochastic fatigue crack growth model proposed by Yang and Manning is employed to analyze the data. The solution of the stochastic differential equation associated with the stochastic model gives us the crack exceedance probability as well as the probability of random time to reach a specified crack size. Through comparison between the analytical and experimental results, it is found the model with a minor modification can fit the experimental data rather well. Once the appropriate stochastic model is established, it can be used for the fatigue reliability prediction of structures made of the tested material. In the present study, in particular, it can be used for the reliability assessment of aging aircraft made of 2024-T351 aluminum alloy.     

Industrial application of RAM modeling: Development and implementation of a RAM simulation model for the Lexan plant at GE Industrial, Plastics

A RAM (reliability, availability and maintenance) model has been built for the GE Industrial, Plastics Lexan^® plant in Bergen op Zoom, The Netherlands. It was based on a Reliability Block Diagram with a Monte Carlo simulation engine. The model has been validated against actual plant data from two different sources, and against local expert opinions, resulting in a satisfactory simulation model. The model was used to assess two key decisions that were (to be) made by GE Industrial, Plastics concerning operation and shutdown policies of the plant. The model results showed that the operation and maintenance could be further improved, and that in doing so the annual production loss could be reduced further.     

Harmonic wavelets based statistical linearization for response evolutionary power spectrum determination

A novel harmonic wavelets based statistical linearization approach is proposed for determining the evolutionary power spectrum (EPS) of the response of nonlinear oscillators subject to stochastic excitation. Specifically, first a mathematically rigorous wavelet-based representation of non-stationary stochastic processes is presented. Next, a representation of the process corresponding to a specific scale and translation level is derived. This procedure leads to an EPS estimation approach which is applicable for estimating not only separable but non-separable in time and frequency EPS as well. Several numerical results are presented in this context. Next, focusing on the case of the stochastic response of a linear system and relying on the orthogonality properties of the developed representation an excitation-response EPS relationship is derived. It is further shown that the excitation-response EPS relationship is valid even for linear time-variant (LTV) systems since the approach possesses inherently the element of time-dependence. Further, an extension via statistical linearization of the input-output EPS relationship for the case of a nonlinear system is developed. The approach involves the concept of assigning optimal and response dependent equivalent stiffness and damping elements corresponding to the specific frequency and time bands. This leads to an iterative determination of the EPS of the system response. Pertinent Monte Carlo simulations demonstrate the reliability and versatility of the approach.     

A stochastic simulation method for uncertainty quantification in the linearized inverse conductivity problem

This paper considers the inverse problem in electrical impedance tomography with non‐informative prior information on the required conductivity function. The problem is approached with a Newton‐type iterative algorithm where the solution of the linearized approximation is estimated using Bayesian inference. The novelty of this work focuses on maximum a posteriori estimation assuming a model that incorporates the linearization error as a random variable. From an analytical expression of this term, we employ Monte Carlo simulation in order to characterize its probability distribution function. This simulation entails sampling an improper prior distribution for which we propose a stable scheme on the basis of QR decomposition. The simulation statistics show that the error on the linearized model is not Gaussian, however, to maintain computational tractability, we derive the posterior probability density function of the solution by imposing a Gaussian kernel approximation to the error density. Numerical results obtained through this approach indicate the superiority of the new model and its respective maximum a posteriori estimator against the conventional one that neglects the impact of the linearization error. Copyright © 2011 John Wiley & Sons, Ltd.     

A Bayesian approach to modeling and predicting pitting flaws in steam generator tubes

Steam generators in nuclear power plants have experienced varying degrees of under-deposit pitting corrosion. A probabilistic model to accurately predict pitting damage is necessary for effective life-cycle management of steam generators. This paper presents an advanced probabilistic model of pitting corrosion characterizing the inherent randomness of the pitting process and measurement uncertainties of the in-service inspection (ISI) data obtained from eddy current (EC) inspections. A Markov chain Monte Carlo simulation-based Bayesian method, enhanced by a data augmentation technique, is developed for estimating the model parameters. The proposed model is able to predict the actual pit number, the actual pit depth as well as the maximum pit depth, which is the main interest of the pitting corrosion model. The study also reveals the significance of inspection uncertainties in the modeling of pitting flaws using the ISI data: Without considering the probability-of-detection issues and measurement errors, the leakage risk resulted from the pitting corrosion would be under-estimated, despite the fact that the actual pit depth would usually be over-estimated.     

A spectral representation based model for Monte Carlo simulation

A new model is proposed for generating samples of real-valued stationary Gaussian processes. The model is based on the spectral representation theorem stating that a weakly stationary process can be viewed as a superposition of harmonics with random properties. The classical use of this theorem for Monte Carlo simulation is based on models consisting of a superposition of harmonics with fixed frequencies but random amplitude and phase. The resulting samples have the same period depending on the discretization of the frequency band. In contrast, the proposed model consists of a superposition of harmonics with random amplitude, phase, and frequency so that different samples have different periods depending on the particular sample values of the harmonic frequencies.

A band limited Gaussian white noise process is used to illustrate the proposed Monte Carlo simulation algorithm and demonstrate that the estimates of the covariance function based on the samples of the proposed model are not periodic.     

New formulation of Cholesky decomposition and applications in stochastic simulation

Monte Carlo simulation plays a significant role in the mechanical and structural analysis due to its versatility and accuracy. Classical spectral representation method is based on the direct decomposition of the power spectral density (PSD) or evolutionary power spectral density (EPSD) matrix through Cholesky decomposition. This direct decomposition of complex matrix usually results in large computational time and storage memory.In this study, a new formulation of the Cholesky decomposition for the EPSD/PSD matrix and corresponding simulation scheme are presented. The key idea to this approach is to separate the phase from the complex EPSD/PSD matrix. The derived real modulus matrix evidently expedites decomposition compared to the direct Cholesky decomposition of the complex EPSD/PSD matrix. In the proposed simulation scheme, the separated phase can be easily assembled. The modulus of EPSD/PSD matrix could be further decomposed into the modulus of coherence matrix (or lagged coherence matrix), which describes the basic coherence structure of stochastic process. The lagged coherence matrix is independence of time and thus remarkably improves the Cholesky decomposition efficiency.The application of the proposed schemes to Gaussian stochastic simulations is presented. Firstly, the previous closed-form wind speed simulation algorithm for equally-spaced locations is extended to a more general situation. Secondly, the proposed approach facilitates the application of interpolation technique in stochastic simulation. The application of interpolation techniques in the wind field simulation is studied as an example.     

A particle code for plasma technological applications

The 2D Direct Simulation Monte Carlo code LasVegas was originally developed and used for the simulation of atmospheric re-entry and the flow around plasma wind tunnel probes. This code is reviewed regarding its applicability to appropriate plasma technological applications. The main DSMC theory and the code structure, features and properties are described in order to classify the LasVegas code in regard to these applications. General restrictions, model uncertainties and geometrical considerations of the DSMC method itself as well as of the LasVegas code put the focus on the vacuum spray process. As proof of concept, results of a supersonic free stream simulation in a vacuum chamber with typical inflow properties are presented and discussed. Appropriate code extensions such as the flow-particle interaction, ionization model and parallel computation capability were identified, thus indicating future numerical works.