Search Results

Design optimization often relies on computational models, which are subjected to a validation process to ensure their accuracy. Because validation of computer models in the entire design space can be costly, we have previously proposed an approach where design optimization and model validation, are concurrently performed using a sequential approach with variable-size local domains. We used test data and statistical bootstrap methods to size each local domain where the prediction model is considered validated and where design optimization is performed. The method proceeds iteratively until the optimum design is obtained. This method however, requires test data to be available in each local domain along the optimization path. In this paper, we refine our methodology by using polynomial regression to predict the size and shape of a local domain at some steps along the optimization process without using test data.

Computer-aided engineering (CAE) is an important tool routinely used to simulate complex engineering systems. Virtual simulations enhance engineering insight into prospective designs and potential design issues and can limit the need for expensive engineering prototypes. For complex engineering systems, however, the effectiveness of virtual simulations is often hindered by excessive computational cost. To minimize the cost of running expensive computer simulations, approximate models of the original model (often called surrogate models or metamodels) can provide sufficient accuracy at a lower computing overhead compared to repeated runs of a full simulation. Metamodel accuracy improves if constructed using space-filling designs of experiments (DOEs). The latter provide a collection of sample points in the design space preferably covering the entire space.

Design optimization often relies on computational models, which are subjected to a validation process to ensure their accuracy. Because validation of computer models in the entire design space can be costly, a recent approach was proposed where design optimization and model validation were concurrently performed using a sequential approach with both fixed and variable-size local domains. The variable-size approach used parametric distributions such as Gaussian to quantify the variability in test data and model predictions, and a maximum likelihood estimation to calibrate the prediction model. Also, a parametric bootstrap method was used to size each local domain. In this article, we generalize the variable-size approach, by not assuming any distribution such as Gaussian. A nonparametric bootstrap methodology is instead used to size the local domains. We expect its generality to be useful in applications where distributional assumptions are difficult to verify, or not met at all.

A simulation-based, system reliability-based design optimization (RBDO) method is presented that can handle problems with multiple failure regions and correlated random variables. Copulas are used to represent the correlation. The method uses a Probabilistic Re-Analysis (PRRA) approach in conjunction with a trust-region optimization approach and local metamodels covering each trust region. PRRA calculates very efficiently the system reliability of a design by performing a single Monte Carlo (MC) simulation per trust region. Although PRRA is based on MC simulation, it calculates “smooth” sensitivity derivatives, allowing therefore, the use of a gradient-based optimizer. The PRRA method is based on importance sampling. It provides accurate results, if the support of the sampling PDF contains the support of the joint PDF of the input random variables. The sequential, trust-region optimization approach satisfies this requirement.

Understanding reliability is critical in design, maintenance and durability analysis of engineering systems. A reliability simulation methodology is presented in this paper for vehicle fleets using limited data. The method can be used to estimate the reliability of non-repairable as well as repairable systems. It can optimally allocate, based on a target system reliability, individual component reliabilities using a multi-objective optimization algorithm. The algorithm establishes a Pareto front that can be used for optimal tradeoff between reliability and the associated cost. The method uses Monte Carlo simulation to estimate the system failure rate and reliability as a function of time. The probability density functions (PDF) of the time between failures for all components of the system are estimated using either limited data or a user-supplied MTBF (mean time between failures) and its coefficient of variation.

A common approach to the validation of simulation models focuses on validation throughout the entire design space. A more recent methodology validates designs as they are generated during a simulation-based optimization process. The latter method relies on validating the simulation model in a sequence of local domains. To improve its computational efficiency, this paper proposes an iterative process, where the size and shape of local domains at the current step are determined from a parametric bootstrap methodology involving maximum likelihood estimators of unknown model parameters from the previous step. Validation is carried out in the local domain at each step. The iterative process continues until the local domain does not change from iteration to iteration during the optimization process ensuring that a converged design optimum has been obtained.

Finite element analysis is a standard tool for deterministic or probabilistic design optimization of dynamic systems. The optimization process requires repeated eigenvalue analyses which can be computationally expensive. Several reanalysis techniques have been proposed to reduce the computational cost including Parametric Reduced Order Modeling (PROM), Combined Approximations (CA), and the Modified Combined Approximations (MCA) method. Although the cost of reanalysis is substantially reduced, it can still be high for models with a large number of degrees of freedom and a large number of design variables. Reanalysis methods use a basis composed of eigenvectors from both the baseline and the modified designs which are in general linearly dependent. To eliminate the linear dependency and improve accuracy, Gram Schmidt orthonormalization is employed which is costly itself.

To improve fuel economy, there is a trend in automotive industry to use light weight, high strength materials. Automotive body structures are composed of several panels which must be downsized to reduce weight. Because this affects NVH (Noise, Vibration and Harshness) performance, engineers are challenged to recover the lost panel stiffness from down-gaging in order to improve the structure borne noise transmitted through the lightweight panels in the frequency range of 100-300 Hz where most of the booming and low medium frequency noise occurs. The loss in performance can be recovered by optimized panel geometry using beading or damping treatment. Topography optimization is a special class of shape optimization for changing sheet metal shapes by introducing beads. A large number of design variables can be handled and the process is easy to setup in commercial codes. However, optimization methods are computationally intensive because of repeated full-order analyses.

Medium trucks constitute a large market segment of the commercial transportation sector, and are also used widely for military tactical operations. Recent technological advances in hybrid powertrains and fuel cell auxiliary power units have enabled design alternatives that can improve fuel economy and reduce emissions dramatically. However, deterministic design optimization of these configurations may yield designs that are optimal with respect to performance but raise concerns regarding the reliability of achieving that performance over lifetime. In this article we identify and quantify uncertainties due to modeling approximations or incomplete information. We then model their propagation using Monte Carlo simulation and perform sensitivity analysis to isolate statistically significant uncertainties. Finally, we formulate and solve a series of reliability-based optimization problems and quantify tradeoffs between optimality and reliability.

Global surrogate modeling aims to build surrogate model with high accuracy in the whole design domain. A major challenge to achieve this objective is how to reduce the number of function evaluations to the original computer simulation model. To date, the most widely used approach for global surrogate modeling is the adaptive surrogate modeling method. It starts with an initial surrogate model, which is then refined adaptively using the mean square error (MSE) or maximizing the minimum distance criteria. It is observed that current methods may not be able to effectively construct a global surrogate model when the underlying black box function is highly nonlinear in only certain regions. A new surrogate modeling method which can allocate more training points in regions with high nonlinearity is needed to overcome this challenge. This article proposes an efficient global surrogate modeling method based on a multi-layer sampling scheme.

This paper presents a methodology to evaluate and optimize discrete event systems, such as an assembly line or a call center. First, the methodology estimates the performance of a system for a single probability distribution of the inputs. Probabilistic Reanalysis (PRRA) uses this information to evaluate the effect of changes in the system configuration on its performance. PRRA is integrated with a program to optimize the system. The proposed methodology is dramatically more efficient than one requiring a new Monte Carlo simulation each time we change the system. We demonstrate the approach on a drilling center and an electronic parts factory.

Monte Carlo simulation is a popular tool for reliability assessment because of its robustness and ease of implementation. A major concern with this method is its computational cost; standard Monte Carlo simulation requires quadrupling the number of replications for halving the standard deviation of the estimated failure probability. Efforts to increase efficiency focus on intelligent sampling procedures and methods for efficient calculation of the performance function of a system. This paper proposes a new method to manage cost that views design as a decision among alternatives with uncertain reliabilities. Information from a simulation has value only if it enables the designer to make a better choice among the alternative options. Consequently, the value of information from the simulation is equal to the gain from using this information to improve the decision. A designer can determine the number of replications that are worth performing by using the method.

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.

Styrene-Butadiene Rubber (SBR), a copolymer of butadiene and styrene, is widely used in the automotive industry due to its high durability and resistance to abrasion, oils and oxidation. Some of the common applications include tires, vibration isolators, and gaskets, among others. This paper characterizes the dynamic behavior of SBR and discusses the suitability of a visco-elastic model of elastomers, known as the Kelvin model, from a mathematical and physical point of view. An optimization algorithm is used to estimate the parameters of the Kelvin model. The resulting model was shown to produce reasonable approximations of measured dynamic stiffness. The model was also used to calculate the self heating of the elastomer due to energy dissipation by the viscous damping components in the model. Developing such a predictive capability is essential in understanding the dynamic behavior of elastomers considering that their dynamic stiffness can in general depend on temperature.

The availability of computational resources has enabled an increased utilization of Design of Experiments (DoE) and metamodeling (response surface generation) for large-scale optimization problems. Despite algorithmic advances however, the analysis of systems such as water jackets of an automotive engine, can be computationally demanding in part due to the required accuracy of metamodels. Because the metamodels may have many inputs, their accuracy depends on the number of training points and how well they cover the entire design (input) space. For this reason, the space-filling properties of the DoE are very important. This paper utilizes a new group-based DoE algorithm with space-filling groups of points to construct a metamodel. Points are added sequentially so that the space-filling properties of the entire group of points is preserved. The addition of points is continuous until a specified metamodel accuracy is met.

Creating a fuel map for simulation of an engine with Variable Valve Actuation (VVA) can be computationally demanding. Design of Experiments (DOE) and metamodeling is one way to address this issue. In this paper, we introduce a sequential process to generate an engine fuel map using Kriging metamodels which account for different engine characteristics such as load and fuel consumption at different operating conditions. The generated map predicts engine output parameters such as fuel rate and load. We first create metamodels to accurately predict the Brake Mean Effective Pressure (BMEP), fuel rate, Residual Gas Fraction (RGF) and CA50 (Crank Angle for 50% Heat Release after top dead center). The last two quantities are used to ensure acceptable combustion. The metamodels are created sequentially to ensure acceptable accuracy is achieved with a small number of simulations.

The internal combustion engine is a source of unwanted vibration on the vehicle body. The unwanted vibration comes from forces on the engine mounts which depend on the engine torque during a transient maneuver. In particular, during a tip-in or a tip-out maneuver, different torque profiles result in different magnitudes of vibration. A desired engine torque shape can be thus obtained to minimize the unwanted vibration. The desired torque shape can be achieved by controlling a set of engine calibration parameters. This paper provides a methodology to determine the spark timing profile to achieve a desired engine torque profile during a tip-out maneuver. The spark timing profiles are described by a third-order polynomial as a function of time. A set of coefficients to define a third-order polynomial (design sites) are first generated using design of experiments (DOE).

Weight reduction is very important in automotive design because of stringent demand on fuel economy. Structural optimization of dynamic systems using finite element (FE) analysis plays an important role in reducing weight while simultaneously delivering a product that meets all functional requirements for durability, crash and NVH. With advancing computer technology, the demand for solving large FE models has grown. Optimization is however costly due to repeated full-order analyses. Reanalysis methods can be used in structural vibrations to reduce the analysis cost from repeated eigenvalue analyses for both deterministic and probabilistic problems. Several reanalysis techniques have been introduced over the years including Parametric Reduced Order Modeling (PROM), Combined Approximations (CA) and the Epsilon algorithm, among others.

Optimizing the trade-off between reliability and cost of operating a microgrid, including vehicles as both loads and sources, can be a challenge. Optimal energy management is crucial to develop strategies to improve the efficiency and reliability of microgrids, as well as new communication networks to support optimal and reliable operation. Prior approaches modeled the grid using MATLAB, but did not include the detailed physics of loads and sources, and therefore missed the transient effects that are present in real-time operation of a microgrid. This article discusses the implementation of a physics-based detailed microgrid model including a diesel generator, wind turbine, photovoltaic array, and utility. All elements are modeled as sources in Simulink. Various loads are also implemented including an asynchronous motor. We show how a central control algorithm optimizes the microgrid by trying to maximize reliability while reducing operational cost.

Reaching a system level reliability target is an inverse problem. Component level reliabilities are determined for a required system level reliability. Because this inverse problem does not have a unique solution, one approach is to tradeoff system reliability with cost and to allow the designer to select a design with a target system reliability, using his/her preferences. In this case, the component reliabilities are readily available from the calculation of the reliability-cost tradeoff. To arrive at the set of solutions to be traded off, one encounters two problems. First, the system reliability calculation is based on repeated system simulations where each system state, indicating which components work and which have failed, is tested to determine if it causes system failure, and second, the task of eliciting and encoding the decision maker's preferences is extremely difficult because of uncertainty in modeling the decision maker's preferences.