• Structural Engineering and Mechanics
• /
• v.38 no.4
• /
• pp.429-441
• /
• 2011

In large-scale problem, a huge size of computational resources is needed for a reliable solution which represents the detailed description of dynamic behavior. Recently, eigenvalue reduction schemes have been considered as important technique to resolve computational resource problems. In addition, the efforts to advance an efficiency of reduction scheme leads to the development of the multi-level system condensation (MLSC) which is initially based on the two-level condensation scheme (TLCS). This scheme was proposed for approximating the lower eigenmodes which represent the global behavior of the structures through the element-level energy estimation. The MLSC combines the multi-level sub-structuring scheme with the previous TLCS for enhancement of efficiency which is related to computer memory and computing time. The present study focuses on the implementation of the MLSC on the direct time response analysis and the frequency response analysis of structural dynamic problems. For the transient time response analysis, the MLSC is combined with the Newmark's time integration scheme. Numerical examples demonstrate the efficiency of the proposed method.

• Journal of the Korea Computer Industry Society
• /
• v.4 no.4
• /
• pp.589-598
• /
• 2003

In the conventional fractal image compression in the DWT(discrete wavelet transform), the domain and range blocks were classified as B${\times}$B block size first before all domain block for each range block was searched. The conventional method has a disadvantages that the encoding time takes too long, since the domain block for entire image was searched. As an enhancement to such inefficiencies and image quality, this paper proposes wavelet-based fractal image compression with multiscale factors. Thus, this proposed method uses multiscale factor along each level and band to enhance an overall image quality. In encoding process of this method, the range blocks are not searched for all the domain blocks; however, using the self affine system the range blocks are selected from the blocks in the upper level. The image qualify of the conventional method is 32.30[dB], and the proposed method is 35.97[dB]. The image quality is increased by 3.67[dB].

• Particle and aerosol research
• /
• v.7 no.3
• /
• pp.71-77
• /
• 2011

Pork belly meat and mackerel are popular meals in Korea. Although a lot of people enjoy cooking these food, there have been some reports that generated organic particles during cooking could be thereas on of lung cancer of nonsmoking housewives. In addition, some experiments show that carcinogens may be included in meat and fishes which we eat usually. For this reason, particle size and concentration in formation during cooking are necessary to figure out the relationship between particles and the diseases. Thus, we identify number concentrations and size distributions of generated nano aerosol in cooking with respect to time, hood operation, and distance between cooking and measurement locations. The maximum concentrations of nano aerosol(diameter sizes are between 10 to 700nm)are decreased after the cooking from $8{\times}10^6{\sharp}/cm^3$ to zeroth order in pork belly meat cooking, and from $3.5{\times}{\times}10^6{\sharp}/cm^3$ to zeroth order in mackerel cooking respectively. When it comes to hood operation during cooking, the detected concentrations of generated aerosols are decreased as in taking flow rate of the hood increases. In cooking pork belly meat, the reduced amount of concentration is about $3{\times}10^6{\sharp}/cm^3$ compared to no hood operation, when hood in taking flow rate is $610m^3/hr$ In mackerel cooking, reduced concentration is $6{\times}10^5{\sharp}/cm^3$ in the same condition. Also, Naphthalene and Fluorene, which are known as polycyclic aromatic hydrocarbons (PAHs), are detected in the generated aerosols during cooking.

• Interaction and multiscale mechanics
• /
• v.4 no.2
• /
• pp.107-122
• /
• 2011

Mechanical behavior in nano-sized structures differs from those in macro sized structures due to surface effect. As the ratio of surface to volume increases, surface effect is not negligible and causes size-dependent mechanical behavior. In order to identify this size effect, atomistic simulations are required; however, it has many limitations because too much computational resource and time are needed. To overcome the restrictions of the atomistic simulations and graft the well-established continuum theories, the continuum model considering surface effect, which is based on the bridging technique between atomistic and continuum simulations, is introduced. Because it reflects the size effect, it is possible to carry out a variety of analysis which is intractable in the atomistic simulations. As a part of the application examples, the homogenization method is applied to micro/nano thin films with porosity and the homogenized elastic coefficients of the nano scale thickness porous films are computed in this paper.

• Smart Structures and Systems
• /
• v.32 no.3
• /
• pp.135-151
• /
• 2023

Using 3D asphalt pavement surface data, a deep and multiscale network named CrackNet-M is proposed in this paper for pixel-level crack detection for improvements in both accuracy and robustness. The CrackNet-M consists of four function-specific architectural modules: a central branch net (CBN), a crack map enhancement (CME) module, three pooling feature pyramids (PFP), and an output layer. The CBN maintains crack boundaries using no pooling reductions throughout all convolutional layers. The CME applies a pooling layer to enhance potential thin cracks for better continuity, consuming no data loss and attenuation when working jointly with CBN. The PFP modules implement direct down-sampling and pyramidal up-sampling with multiscale contexts specifically for the detection of thick cracks and exclusion of non-crack patterns. Finally, the output layer is optimized with a skip layer supervision technique proposed to further improve the network performance. Compared with traditional supervisions, the skip layer supervision brings about not only significant performance gains with respect to both accuracy and robustness but a faster convergence rate. CrackNet-M was trained on a total of 2,500 pixel-wise annotated 3D pavement images and finely scaled with another 200 images with full considerations on accuracy and efficiency. CrackNet-M can potentially achieve crack detection in real-time with a processing speed of 40 ms/image. The experimental results on 500 testing images demonstrate that CrackNet-M can effectively detect both thick and thin cracks from various pavement surfaces with a high level of Precision (94.28%), Recall (93.89%), and F-measure (94.04%). In addition, the proposed CrackNet-M compares favorably to other well-developed networks with respect to the detection of thin cracks as well as the removal of shoulder drop-offs.

• Interaction and multiscale mechanics
• /
• v.4 no.3
• /
• pp.219-237
• /
• 2011

Recent advances in scientific computing enable the full atomistic simulation of DNA molecules. However, there exists length and time scale limitations in molecular dynamics (MD) simulation for large DNA molecules. In this work, a two-level homogenization of DNA molecules is proposed. A wavelet projection method is first introduced to form a coarse-grained DNA molecule represented with superatoms. The coarsened MD model offers a simplified molecular structure for the continuum description of DNA molecules. The coarsened DNA molecular structure is then homogenized into a three-dimensional beam with embedded molecular properties. The methods to determine the elasticity constants in the continuum model are also presented. The proposed continuum model is adopted for the study of mechanical behavior of DNA loop.

• Interaction and multiscale mechanics
• /
• v.1 no.1
• /
• pp.83-101
• /
• 2008

A derivation is given of two-scale models that are able to describe deformation and flow in a fluid-saturated and progressively fracturing porous medium. From the micromechanics of the flow in the cavity, identities are derived that couple the local momentum and the mass balances to the governing equations for a fluid-saturated porous medium, which are assumed to hold on the macroscopic scale. By exploiting the partition-of-unity property of the finite element shape functions, the position and direction of the fractures are independent from the underlying discretization. The finite element equations are derived for this two-scale approach and integrated over time. The resulting discrete equations are nonlinear due to the cohesive crack model and the nonlinearity of the coupling terms. A consistent linearization is given for use within a Newton-Raphson iterative procedure. Finally, examples are given to show the versatility and the efficiency of the approach.

• Journal of Mechanical Science and Technology
• /
• v.20 no.12
• /
• pp.2147-2158
• /
• 2006

The continuous wavelet transform (CWT) has a frequency-adaptive time-frequency tiling property, which makes it popular for the analysis of dispersive elastic wave signals. However, because the time-frequency tiling of CWT is not signal-dependent, it still has some limitations in the analysis of elastic waves with spectral components that are dispersed rapidly in time. The objective of this paper is to introduce an advanced time-frequency analysis method, called the dispersion-based continuous wavelet transform (D-CWT) whose time-frequency tiling is adaptively varied according to the dispersion relation of the waves to be analyzed. In the D-CWT method, time-frequency tiling can have frequency-adaptive characteristics like CWT and adaptively rotate in the time-frequency plane depending on the local wave dispersion. Therefore, D-CWT provides higher time-frequency localization than the conventional CWT. In this work, D-CWT method is applied to the analysis of dispersive elastic waves measured in waveguide experiments and an efficient procedure to extract information on the dispersion relation hidden in a wave signal is presented. In addition, the ridge property of the present transform is investigated theoretically to show its effectiveness in analyzing highly time-varying signals. Numerical simulations and experimental results are presented to show the effectiveness of the present method.

• Journal of Astronomy and Space Sciences
• /
• v.39 no.3
• /
• pp.79-86
• /
• 2022

Thermospheric wind observations from high to mid latitudes are compared with the newly developed Multiscale Atmosphere Geospace Environment (MAGE) model for the Nov 3-4 geomagnetic storm. The observation and simulation comparison shows a very good agreement and is better at high latitudes in general. We were able to identify a thermospheric poleward wind reduction possibly linked to a northward turning of the Interplanetary Magnetic Field (IMF) at ~22 UT on Nov 3 and an enhancement of the poleward wind to a southward turning near 10 UT on Nov 4 at high latitudes. An IMF southward turning may have led to an enhancement of equatorward winds at Boulder, Colorado near midnight. Simultaneous occurrence of aurora may be associated with an IMF By turning negative. The MAGE model wind simulations are consistent with observations in these cases. The results show the model can be a very useful tool to further study the magnetosphere and ionosphere coupling on short time scales.

• Interaction and multiscale mechanics
• /
• v.3 no.2
• /
• pp.111-121
• /
• 2010

Dynamic analysis of nano/micro bio-sensors based on a multiscale atomistic/continuum theory is introduced. We use a generalized atomistic finite element method (GAFEM) to analyze a bio-sensor which has $3{\times}N_a{\times}N_p$ degrees of freedom, where $N_p$ is the number of representative unit cells and $N_a$ is the number of atoms per unit cell. The stiffness matrix is derived from interatomic potential between pairs of atoms. This work contains two studies: (1) the resonance analysis of nano bio-sensors with different amount of target analyte and (2) the dependence of resonance frequency on finite element mesh. We also examine the Courant-Friedrichs-Lewy (CFL) condition based on the highest resonance frequency. The CFL condition is the criterion for the time step used in the dynamic analysis by GAFEM. Our studies can be utilized to predict the performance of micro/nano bio-sensors from atomistic perspective.