Search Results

Our recent work has shown that representation of systems using a reliability block diagram can be used as a decision making tool. In decision making, we called these block diagrams decision topologies. In this paper, we generalize the results and show that decision topologies can be used to make many engineering decisions and can in fact replace decision analysis for most decisions. We also provide a meta-proof that the proposed method using decision topologies is entirely consistent with decision analysis at the limit. The main advantages of the method are that (1) it provides a visual representation of a decision situation, (2) it can easily model tradeoffs, (3) it can incorporate binary attributes, (4) it can model preferences with limited information, and (5) it can be used in a low-fidelity sense to quickly make a decision.

We present an Accelerated Life Testing (ALT) methodology along with a design for fatigue approach, using Gaussian or non-Gaussian excitations. The accuracy of fatigue life prediction at nominal loading conditions is affected by model and material uncertainty. This uncertainty is reduced by performing tests at a higher loading level, resulting in a reduction in test duration. Based on the data obtained from experiments, we formulate an optimization problem to calculate the Maximum Likelihood Estimator (MLE) values of the uncertain model parameters. In our proposed ALT method, we lift all the assumptions on the type of life distribution or the stress-life relationship and we use Saddlepoint Approximation (SPA) method to calculate the fatigue life Probability Density Functions (PDFs).

A reliability analysis method is presented for time-dependent systems under uncertainty. A level-crossing problem is considered where the system fails if its maximum response exceeds a specified threshold. The proposed method uses a double-loop optimization algorithm. The inner loop calculates the maximum response in time for a given set of random variables, and transforms a time-dependent problem into a time-independent one. A time integration method is used to calculate the response at discrete times. For each sample function of the response random process, the maximum response is found using a global-local search method consisting of a genetic algorithm (GA), and a gradient-based optimizer. This dynamic response usually exhibits multiple peaks and crosses the allowable response level to form a set of complex limit states, which lead to multiple most probable points (MPPs).

Reliability-based design optimization accounts for variation. However, it assumes that statistical information is available in the form of fully defined probabilistic distributions. This is not true for a variety of engineering problems where uncertainty is usually given in terms of interval ranges. In this case, interval analysis or possibility theory can be used instead of probability theory. This paper shows how possibility theory can be used in design and presents a computationally efficient sequential optimization algorithm. The algorithm handles problems with only uncertain or a combination of random and uncertain design variables and parameters. It consists of a sequence of cycles composed of a deterministic design optimization followed by a set of worst-case reliability evaluation loops. A crank-slider mechanism example demonstrates the accuracy and efficiency of the proposed sequential algorithm.

A Reliability-Based Design Optimization (RBDO) method for multiple failure regions is presented. The method uses a Probabilistic Re-Analysis (PRRA) approach in conjunction with an approximate global metamodel with local refinements. The latter serves as an indicator to determine the failure and safe regions. PRRA calculates very efficiently the system reliability of a design by performing a single Monte Carlo (MC) simulation. Although PRRA is based on MC simulation, it calculates “smooth” sensitivity derivatives, allowing therefore, the use of a gradient-based optimizer. An “accurate-on-demand” metamodel is used in the PRRA that allows us to handle problems with multiple disjoint failure regions and potentially multiple most-probable points (MPP). The multiple failure regions are identified by using a clustering technique. A maximin “space-filling” sampling technique is used to construct the metamodel. A vibration absorber example highlights the potential of the proposed method.

Deep charge and discharge cycling of 24 Volt battery strings composed of two 12 Volt VRLA batteries wired in series affects reliability and life expectancy. This is a matter of interest in vehicle power source applications. These cycles include those specific operational cases requiring the delivery of the full storage capacity during discharge. The concern here is related to applications where batteries serve as a primary power source and the energy content is an issue. It is a common practice for deep cycling a 24 volt battery string to simply add the specified limit voltages during charge and discharge for the individual 12 Volt batteries. In reality, the 12 Volt batteries have an inherent capacity variability and are not identical in their performance characteristics. The actual voltages of the individual 12 Volt batteries are not identical.

Existing methods for design optimization under uncertainty assume that a high level of information is available, typically in the form of data. In reality, however, insufficient data prevents correct inference of probability distributions, membership functions, or interval ranges. In this article we use an engine design example to show that optimal design decisions and reliability estimations depend strongly on uncertainty characterization. We contrast the reliability-based optimal designs to the ones obtained using worst-case optimization, and ask the question of how to obtain non-conservative designs with incomplete uncertainty information. We propose an answer to this question through the use of Bayesian statistics. We estimate the truck's engine reliability based only on available samples, and demonstrate that the accuracy of our estimates increases as more samples become available.

Early in the engineering design cycle, it is difficult to quantify product reliability due to insufficient data or information to model uncertainties. Probability theory can not be therefore, used. Design decisions are usually based on fuzzy information which is imprecise and incomplete. Various design methods such as Possibility-Based Design Optimization (PBDO) and Evidence-Based Design Optimization (EBDO) have been developed to systematically treat design with non-probabilistic uncertainties. In practical engineering applications, information regarding the uncertain variables and parameters may exist in the form of sample points, and uncertainties with sufficient and insufficient information may exist simultaneously. Most of the existing optimal design methods under uncertainty can not handle this form of incomplete information. They have to either discard some valuable information or postulate the existence of additional information.

In today's light-weight vehicles, the strength of spot welds plays an important role in overall product integrity, reliability and customer satisfaction. Naturally, there is a need for a quick and reliable technique to inspect the quality of the welds. In the past, the primary quality control tests for detecting weld defects are the destructive chisel test and peel test [1]. The non-destructive evaluation (NDE) method currently used in industry is based on ultrasonic inspection [2, 3, 4]. The technique is not always successful in evaluating the nugget size, nor is it effective in detecting the so-called “cold” or “stick” welds. Therefore, it is necessary to develop a precise and reliable noncontact NDE method for spot welds. There have been numerous studies in predicting the weld nugget size by considering the spot-weld process [5, 6].

Multicore microcontrollers are rapidly making their way into the automotive industry. We have adopted the Cilk approach (MIT 1994) to develop a pure ANSI C Fork-Join dynamic scheduling runtime middle-layer with a work-stealing scheduler targeted for automotive multicore embedded systems. This middle-layer could be running on top of any AUTOSAR compliant multicore RTOS. We recently have successfully integrated our runtime layer into parts of legacy Ford powertrain software at Ford Motor Company. We have used the 3-core AURIX multicore chip from Infineon and the multicore RTA-OS. For testing purposes, we have forked some parallelizable functions inside two periodic tasks in Ford legacy powertrain software to be dynamically scheduled and executed on the available cores. Our preliminary evaluation showed 1.3–1.4x speedups for these two forked tasks.

In reliability design, often, there is scarce data for constructing probabilistic models. It is particularly challenging to model uncertainty in variables when the type of their probability distribution is unknown. Moreover, it is expensive to estimate the upper and lower bounds of the reliability of a system involving such variables. A method for modeling uncertainty by using Polynomial Chaos Expansion is presented. The method requires specifying bounds for statistical summaries such as the first four moments and credible intervals. A constrained optimization problem, in which decision variables are the coefficients of the Polynomial Chaos Expansion approximation, is formulated and solved in order to estimate the minimum and maximum values of a system’s reliability. This problem is solved efficiently by employing a probabilistic re-analysis approach to approximate the system reliability as a function of the moments of the random variables.

Importance Sampling is a popular method for reliability assessment. Although it is significantly more efficient than standard Monte Carlo simulation if a suitable sampling distribution is used, in many design problems it is too expensive. The authors have previously proposed a method to manage the computational cost in standard Monte Carlo simulation that views design as a choice among alternatives with uncertain reliabilities. Information from simulation has value only if it helps the designer make a better choice among the alternatives. This paper extends their method to Importance Sampling. First, the designer estimates the prior probability density functions of the reliabilities of the alternative designs and calculates the expected utility of the choice of the best design. Subsequently, the designer estimates the likelihood function of the probability of failure by performing an initial simulation with Importance Sampling.

In the automotive industry, in order to characterize and evaluate damping treatments, it is a common practice to employ Oberst bar tests as specified by ASTM E756 and SAE J1637. The ASTM standard provides equations for sandwiched Oberst bars. These equations allow engineers to extract the properties of the visco-elastic core. For certain type of automotive constrained-layer damping treatments, such as the Aluminum Foilback, it is often convenient and desirable to prepare the Oberst bar samples with production-intent configuration. Unfortunately, these configurations are often asymmetric. Therefore, the composite Oberst bar data cannot be post-processed by employing the ASTM equations. In this study, the ASTM equations for sandwiched bars are modified to accommodate for asymmetric Oberst bar configurations. The finite element method is used to validate the derived equations by performing a “Virtual Oberst Bar test.”

A new second-order Saddlepoint Approximation (SA) method for structural reliability analysis is introduced. The Mean-value Second-order Saddlepoint Approximation (MVSOSA) is presented as an extension to the Mean-value First-order Saddlepoint Approximation (MVFOSA). The proposed method is based on a second-order Taylor expansion of the limit state function around the mean value of the input random variables. It requires not only the first but also the second-order sensitivity derivatives of the limit state function. If sensitivity analysis must be avoided because of computational cost, a quadrature integration approach, based on sparse grids, is also presented and linked to the saddlepoint approximation (SGSA - Sparse Grid Saddlepoint Approximation). The SGSA method is compared with the first and second-order SA methods in terms of accuracy and efficiency. The proposed MVSOSA and SGSA methods are used in the reliability analysis of two examples.

Automotive electronics and software is getting complex day by day. More and more features and functions are offered and supported by software in place of hardware. Communication is carried out on the CAN bus instead of hard wired circuits. This architectural transition facilitates lots of flexibility, agility and economy in development. However, it introduces risk of unexpected failures due to insufficient testing and million of possible combinations, which can be created by users during the life time of a product. Model based development supports an effective way of handling these complexities during simulation and also provide oracle for its validation. Based on priorities and type of applications, test vectors can be auto generated and can be used for formal verification of the models. These auto-generated test vectors are valuable assets in testing and can be effectively reused for target hardware (ECU) verification.

The system response of many engineering systems depends on time. A random process approach is therefore, needed to quantify variation or uncertainty. The system input may consist of a combination of random variables and random processes. In this case, a time-dependent reliability analysis must be performed to calculate the probability of failure within a specified time interval. This is known as cumulative probability of failure which is in general, different from the instantaneous probability of failure. Failure occurs if the limit state function becomes negative at least at one instance within a specified time interval. Time-dependent reliability problems appear if for example, the material properties deteriorate in time or if random loading is involved which is modeled by a random process. Existing methods to calculate the cumulative probability of failure provide an upper bound which may grossly overestimate the true value.

An approach for Probabilistic Reanalysis (PRA) of a system is presented. PRA calculates very efficiently the system reliability or the average value of an attribute of a design for many probability distributions of the input variables, by performing a single Monte Carlo simulation. In addition, PRA calculates the sensitivity derivatives of the reliability to the parameters of the probability distributions. The approach is useful for analysis problems where reliability bounds need to be calculated because the probability distribution of the input variables is uncertain or for design problems where the design variables are random. The accuracy and efficiency of PRA is demonstrated on vibration analysis of a car and on system reliability-based optimization (RBDO) of an internal combustion engine.

This paper proposes an approach for optimal design of multilevel systems under uncertainty. The approach utilizes the stochastic extension of the analytical target cascading formulation. The reliability of satisfying the probabilistic constraints is computed by means of the most probable point method using the hybrid mean value algorithm. A linearization technique is employed for estimating the propagation of uncertainties throughout the problem hierarchy. The proposed methodology is applied to a piston-ring/cylinder-liner engine subassembly design problem. Specifically, we assess the impact of variations in manufacturing-related properties such as surface roughness on engine attributes such as brake-specific fuel consumption. Results are compared to the ones obtained using Monte Carlo simulation.

In this paper, the development of random vibration testing schedules for durability design verification of engine mounted products is presented, based on the equivalent fatigue damage concept and the 95th-percentile customer engine usage data for 150,000 miles. Development of the 95th-percentile customer usage profile is first discussed. Following that, the field engine excitation and engine duty cycle definition is introduced. By using a simplified transfer function of a single degree-of-freedom (SDOF) system subjected to a base excitation, the response acceleration and stress PSDs are related to the input excitation in PSD, which is the equivalent fatigue damage concept. Also, the narrow-band fatigue damage spectrum (FDS) is calculated in terms of the input excitation PSD based on the Miner linear damage rule, the Rayleigh statistical distribution for stress amplitude, a material's S-N curve, and the Miles approximate solution.

An accurate and efficient Monte Carlo simulation (MCS) method is developed in this paper for limit state-based reliability analysis, especially at system levels, by using a response surface approximation of the failure indicator function. The Moving Least Squares (MLS) method is used to construct the response surface of the indicator function, along with an Optimum Symmetric Latin Hypercube (OSLH) as the sampling technique. Similar to MCS, the proposed method can easily handle implicit, highly nonlinear limit-state functions, with variables of any statistical distributions and correlations. However, the efficiency of MCS can be greatly improved. The method appears to be particularly efficient for multiple limit state and multiple design point problem. A mathematical example and a practical example are used to highlight the superior accuracy and efficiency of the proposed method over traditional reliability methods.