Seung-Kyum Choi

Polynomial Chaos Expansion with Latin Hypercube Sampling for Estimating Response Variability

A computationally efficient procedure for quantifying uncertainty and finding significant parameters of uncertainty models is presented. To deal with the random nature of input parameters of structural models, several efficient probabilistic methods are investigated. Specifically, the polynomial chaos expansion with Latin hypercube sampling is used to represent the response of an uncertain system. Latin hypercube sampling is employed for evaluating the generalized Fourier coefficients of the polynomial chaos expansion. Because the key challenge in uncertainty analysis is to find the most significant components that drive response variability, analysis of variance is employed to find the significant parameters of the approximation model. Several analytical examples and a large finite element model of a joined-wing are used to verify the effectiveness of this procedure.

Design and fabrication of periodic lattice-based cellular structures

ABSTRACTA methodology, which consists of design, optimization and evaluation of periodic lattice-based cellular structures fabricated by additive manufacturing, is presented. A user-friendly design framework for lattice cellular structures is developed by using a size optimization algorithm. A 3D modeling process for the lattice-based cellular structures is introduced for non-linear finite element analysis and production. The approach is demonstrated on compression block with periodic lattice-based unit cells. First, based on loading condition, most appropriate lattice layout is selected. Then, for the selected lattice layout, the lattice components are modeled as simple beam and size of the beam cross sections is optimized using in-house optimization approach for both yield and local buckling criteria. The 3D model for the optimized lattice structure is built and non-linear finite element study is conducted to predict the performance. Physical parts are 3D printed and tested to compare with the simulatio...

Estimation of structural reliability for Gaussian random fields

This research develops a stochastic analysis procedure for Gaussian random fields in structural reliability estimation using an orthogonal transform and a stochastic expansion with Latin Hypercube sampling. The efficiency of the current simulation procedure is achieved by combination of the Karhunen – Loeve transform with stochastic analysis of polynomial chaos expansion. The Karhunen – Loeve transform enables generation of random fields within the framework of Latin Hypercube sampling and dimensionality reduction of the random variables. The polynomial chaos expansion can reduce computational effort of uncertainty quantification in highly nonlinear engineering design applications. In order to show the applicability of the method, the material properties of a cantilever plate and a supercavitating torpedo are treated as random fields.

Simulation-Based Robust Design of Multiscale Products

The current research proposes an integrated framework for product design that incorporates simulation-based tools into the early design stage to achieve optimum multiscale systems. The method to determine the appropriate cellular material-property relations for the internal material structures of the system is through a topology optimization technique and a multiscale design process. Specifically, the reliability-based topology optimization (RBTO) and a multi-attribute decision design method are integrated into the inductive design exploration method (IDEM). The RBTO method is introduced to determine optimal topologies at the mesoscale. The multi-attribute decision design method is used for decision support in the design process of the macroscale systems. IDEM offers the capability for concurrent design on multiple scales providing an approach for the integration of the other two methods. An example of the developed multiscale design framework is presented in terms of a hydrogen storage tank used in a hydrogen fuel cell for automotive application. The multiscale tank design will feature a high strength cellular structured wall, resulting in a large weight reduction.

Robust design of mechanical systems via stochastic expansion

This paper discusses stochastic optimisation using Polynomial Chaos Expansion (PCE) with Latin Hypercube Sampling (LHS). PCE provides a means to quantify the uncertainty of highly non-linear structural models involving inherent randomness of geometric, material, and loading properties. PCE, specifically the non-intrusive formulation, is used to construct surrogates of stochastic responses for optimisation procedures. A standard optimisation algorithm is then used. In particular, the material properties of structural systems are assumed to be uncertain and treated as uncorrelated random variables. Implementation of the method is demonstrated for a three-bar structure and a complex engineering structure of an uninhabited joined-wing aircraft.

A Multilevel Upscaling Method for Material Characterization of Additively Manufactured Part Under Uncertainties

An integrated multiscale modeling framework that incorporates a simulation-based upscaling technique is developed and implemented for the material characterization of additively manufactured cellular structures in this paper. The proposed upscaling procedure enables the determination of homogenized parameters at multiple levels by matching the probabilistic performance between fine and coarse scale models. Polynomial chaos expansion (PCE) is employed in the upscaling procedure to handle the computational burden caused by the input uncertainties. Efficient uncertainty quantification is achieved at the mesoscale level by utilizing the developed upscaling technique. The homogenized parameters of mesostructures are utilized again at the macroscale level in the upscaling procedure to accurately obtain the overall material properties of the target cellular structure. Actual experimental results of additively manufactured parts are integrated into the developed procedure to demonstrate the efficacy of the method.

Extracting Consumer Preference From User-Generated Content Sources Using Classification

The use of online, user-generated content for consumer preference modeling has been a recent topic of interest among the engineering and marketing communities. With the rapid growth of many different types of user-generate content sources, the tasks of reliable opinion extraction and data interpretation are critical challenges. This research investigates one of the largest and most-active content sources, Twitter, and its viability as a content source for preference modeling. Support Vector Machine (SVM) is used for sentiment classification of the messages, and a Twitter query strategy is developed to categorize messages according to product attributes and attribute levels. Over 7,000 messages are collected for a smartphone design case study. The preference modeling results are compared with those from a typical product review study, including over 2,500 product reviews. Overall, the results demonstrate that consumers do express their product opinions through Twitter; thus, this content source could potentially facilitate product design and decision-making via preference modeling.Copyright

Remotely controlled laboratory experiments: Creation and examples

Most users who can only connect to their university through distance learning enabled programs have no other choice than to sit out the experimental side of education. Remote Labs have the greatest potential to overcome the bottleneck in distance education. The goal of Remote Laboratory implementation is to grant these students access to laboratory equipment. Although there is not currently a way to perfectly emulate these encounters completely, there are many practices and tools that will help match a traditional kinesthetic environment in a Remote Lab. An experiment was created during thesis research to obtain experimental data and analyze the ability of Remote Labs to be integrated with current coursework. Surveys were distributed to appraise the perception of the lab. The collected data indicated that the perceptions a student carries about the effectiveness of Remote Laboratories improves after they perform the experiment.

A sequential multi-fidelity metamodeling approach for data regression

Abstract Multi-fidelity (MF) metamodeling approaches have attracted significant attention recently for data regression because they can make a trade-off between high accuracy and low computational expense by integrating the information from high-fidelity (HF) and low-fidelity (LF) models. To facilitate the usage of the MF metamodeling approaches, there are still challenging issues on the sample size ratio between HF and LF models and the locations of samples since these two components have profound effects on the prediction accuracy of the MF metamodels. In this study, a sequential multi-fidelity (SMF) metamodeling approach is proposed to address the issues of 1) where to allocate the LF and HF sample points, and 2) how to obtain an optimal combination of the high and low-fidelity sample sizes for a given computational budget and a high-to-low simulation cost ratio. Firstly, sequential objective formulations, with the objective to reduce the estimation of prediction error of MF metamodel, are constructed to update the LF and HF sampling data. Secondly, a decision criterion is proposed to determine whether one HF experiment or several LF experiments with the equivalent computational cost should be selected to update the MF metamodel. The proposed criterion is developed according to which selection will have a greater potential value to improve the prediction accuracy of the MF metamodel. To demonstrate the effectiveness and merits of the proposed SMF metamodeling approach, two numerical examples and a practical aerospace application example are used. Results show that the proposed approach can generate more accurate MF metamodels by providing the optimal high-to-low sample size ratio and sample locations.

Optimal Synthesis of Mesostructured Materials under Uncertainty

Uncertainties in material properties, geometry, manufacturing processes, and operational environments are clearly critical at all scales (nano-, micro-, meso-, and macro-scale). Specifically, the incorporation of reliability analysis is critical in designing mesostructured materials when material properties are uncertain. The concept of mesostructured materials is motivated by the desire to put material only where it is needed for a specific application. This research develops a reliability-based synthesis method to design mesostructured materials under uncertainty, which have superior structural compliant performance per weight than parts with bulk material or foams. The efficiency of the proposed framework is achieved with the combination of topology optimization and stochastic approximation which utilizes stochastic local regression and Latin Hypercube sampling. The effectiveness of the proposed framework is demonstrated with two examples including a ground truss structure and an automotive component which is composed of mesostructured material. Nomenclature Ai = area of cross section of each truss element di