• 한국CDE학회논문집
• /
• 제11권2호
• /
• pp.128-137
• /
• 2006

In this paper, a CAD/CAE integrated optimal design system is developed, in which design and analysis process is automated using CAD/CAE softwares for a complex model in which the modeling by parametric feature is not easy to apply. Unigraphics is used for CAD modeling, in which the process is automated by using UG/Knowledge Fusion for modeling itself and UG/Open API function for the other functions respectively. Structural analyses are also carried out automatically by ANSYS using the imported parasolid model. The developed system is applied for the PLS(Plasma Lighting System) consisting of more than 20 components, which is a next generation illumination system that is used to illuminate stadium or outdoor advertizing panel. The analyses include responses by static, wind and impact loads. As a result of analyses, tilt assembly, which is a link between upper and lower body, is found to be the most critical component bearing higher stresses. Experiment is conducted using MTS to validate the analysis result. Optimization is carried out using the software Visual DOC for the tilt assembly to minimize material volume while maintaining allowable stress level. As a result of optimization, the maximum stress is reduced by 57% from the existing design, though the material volume has increased by 21%.

• 한국진공학회:학술대회논문집
• /
• 한국진공학회 2012년도 제43회 하계 정기 학술대회 초록집
• /
• pp.267-267
• /
• 2012

최근 국내 반도체 장비 업체들에 의해서 차세대 반도체용 450 mm 웨이퍼 공정용 장비 개발이 진행 중에 있다. 반도체 산업은 계속해서 반도체 칩의 크기를 작게 하고, 웨이퍼 크기를 늘리면서 웨이퍼 당 칩수를 증가시켜 생산성을 향상해오고 있다. 현재 300 mm 웨이퍼에서 450 mm 웨이퍼를 도입하게 되면, 생산성 뿐만 아니라 30%의 비용절감과 50%의 cycle-time 단축이 기대되고 있다. 장비에 대한 이해와 공정에 대한 해석 능력을 위해 비용과 시간이 많이 들기 때문에 최근 컴퓨터를 활용한 수치 모델링이 진행되고 있다. 또한, 수치 모델링은 실험 결과와의 비교가 필수적이다. 본 연구에서는 450 mm 웨이퍼 공정용 장비의 전자밀도를 cut off probe를 통해 100 mTorr에 서 Ar 플라즈마를 파워에 따라 측정했다. 13.56 MHz 200 W, 500 W, 1,000 W로 입력 파워가 증가하면서 웨이퍼 중심에서 $6.0{\times}10^9#/cm^3$, $1.35{\times}10^{10}#/cm^3$, $2.4{\times}10^{10}#/cm^3$로 증가했다. 450 mm 웨이퍼 영역에서 전자 밀도의 불균일도는 각각 10.31%, 3.24%, 4.81% 였다. 또한, 이 450 mm 웨이퍼용 CCP 장비를 축대칭 2차원으로 형상화하고, 전극에 13.56 MHz를 직렬로 연결된 blocking capacitor ($1{\times}10^{-6}$ F/$m^2$)를 통해 인가할 수 있도록 상용 유체 모델 소프트웨어(CFD-ACE+, EXI corp)를 이용하여 계산하였다. 주요 전자-중성 충돌 반응으로 momentum transfer, ionization, excitation, two-step ionization을 고려했고, $Ar^+$와 $Ar^*$의 표면 재결합 반응은 sticking coefficient를 1로 가정했다. CFD-ACE+의 CCP 모델을 통해 Poisson 방정식을 풀어서 sheath와 wave effect를 고려하였다. Stochastic heating을 고려하지 않았을 때, 플라즈마 흡수 파워가 80 W, 160 W, 240 W에서 실험 투입 전력 200 W, 500 W, 1,000 W일 때와 유사한 반경 방향의 플라즈마 밀도 분포를 보였다. 200 W, 500 W, 1,000 W일 때의 전자밀도 분포는 수치 모델링과 전 범위에서 각각 10%, 3%, 2%의 오차를 보였다. 450 mm의 전극에 13.56 MHz의 전력을 인가할 때, 파워가 증가할수록 전자밀도의 최대값의 위치가 웨이퍼 edge에서 중심으로 이동하고 있음을 실험과 모델링을 통해 확인할 수 있었다.

• 전자공학회논문지
• /
• 제52권3호
• /
• pp.105-115
• /
• 2015

반도체 식각 전산모사에서는 플라즈마 입자를 생성하는 소스의 모델링이 필요하다. 본 논문에서는 플라즈마 식각 공정에서 사용하는 소스를 확률분포함수로 모델링하고, 몬테칼를로 방법을 이용하여 특정 프로프일의 플럭스를 계산하는 실험을 하였다. 소스의 모델링 파라미터로 소스와 셀 사이의 거리, 소스에서 방사하는 입자수가 있고, 플럭스 계산에 미치는 추가적인 파라미터로 프로파일 상의 셀의 수(셀의 면적)이 있다. 방사하는 입자 분포는 사용하는 소스의 물성에 따라 가우시안 분포와 코사인 분포로 모델링 할 수 있는데, 본 논문은 이들 각각에 대하여 파라미터를 바꿔가며 전산모사를 한 결과를 보인다. 오차율은 가우지안(Incident Flux)과 코사인분포(Incident Neutral Flux)에서 모두 입자 수의 증가에 따라 상당부분 감소하였으나 처리시간은 이보다 더 증가하였다. 셀수와 거리의 증가는 오차율을 약간 증가시켰고 처리시간도 증가시켰다. 본 논문의 실험 결과를 통해 처리 시간을 고려하여 적합한 플럭스의 계산을 유추할 수 있다.

• 천문학회지
• /
• 제46권1호
• /
• pp.49-63
• /
• 2013

Nonthermal radiation from supernova remnants (SNRs) provides observational evidence and constraints on the diffusive shock acceleration (DSA) hypothesis for the origins of Galactic cosmic rays (CRs). Recently it has been recognized that a variety of plasma wave-particle interactions operate at astrophysical shocks and the detailed outcomes of DSA are governed by their complex and nonlinear interrelationships. Here we calculate the energy spectra of CR protons and electrons accelerated at Type Ia SNRs, using time-dependent, DSA simulations with phenomenological models for magnetic field amplification due to CR streaming instabilities, Alf$\acute{e}$enic drift, and free escape boundary. We show that, if scattering centers drift with the Alf$\acute{e}$en speed in the amplified magnetic fields, the CR energy spectrum is steepened and the acceleration efficiency is significantly reduced at strong CR modified SNR shocks. Even with fast Afv$\acute{e}$nic drift, DSA can still be efficient enough to develop a substantial shock precursor due to CR pressure feedback and convert about 20-30% of the SN explosion energy into CRs. Since the high energy end of the CR proton spectrum is composed of the particles that are injected in the early stages, in order to predict nonthermal emissions, especially in X-ray and ${\gamma}-ray$ bands, it is important to follow the time dependent evolution of the shock dynamics, CR injection process, magnetic field amplification, and particle escape. Thus it is crucial to understand the details of these plasma interactions associated with collisionless shocks in successful modeling of nonlinear DSA.

• Journal of Advanced Marine Engineering and Technology
• /
• 제31권3호
• /
• pp.282-292
• /
• 2007

The present paper describes the results of high speed photography, acoustic emission (AE) detection and plasma light emission (LE) measurement during $CO_2$ laser welding of 304 stainless steel in different processing conditions. Video images with high spatial and temporal resolution allowed to observe the melt dynamics and keyhole evolution. The existence of keyhole was confirmed by the slag motion on the weld pool. The characteristic frequencies of flow instability and keyhole fluctuations at different welding speed were measured and compared with the results of Fourier analyses of temporal AE and LE spectra. The experimental results were compared with the newly developed numerical model of keyhole dynamics. The model is based on the assumption that the propagation of front part of keyhole into material is due to the melt ejection driven by laser induced surface evaporation. The calculations predict that a high speed melt flow is induced at the front part of keyhole when the sample travel speed exceeds several 10 mm/s. The numerical analysis also shows the hump formation on the front keyhole wall surface. Experimentally observed melt behavior and transformation of the AE and LE spectra with variation of welding speed are qualitatively in good agreement with the model predictions.

• Journal of Electrical Engineering and Technology
• /
• 제10권5호
• /
• pp.2046-2051
• /
• 2015

Currently, most high-voltage gas circuit breakers (CBs) include asymmetrical geometries in the shield, the tank, the hot-gas exhaust, and the connection parts for bushings. For this reason, a 3-dimensional (3-D) analysis of the insulation capability is necessary, rather than a 2-D analysis. However, a 3-D analysis has difficulties due to the computational time and complex modeling. This paper presents a 3-D analysis considering the asymmetry in high-voltage gas CBs and a technique to reduce the calculation time. In the proposed technique, the arc plasma requiring the most computational time is first calculated by a 2-D analysis. Then, the results such as pressure, temperature, and velocity are input as a source for the 3-D analysis. This technique is applied to a 145kV self-blast-type CB and the analysis result exhibits good agreement with the experimental result.

• Nuclear Engineering and Technology
• /
• 제52권2호
• /
• pp.287-295
• /
• 2020

Group method of data handling (GMDH) is considered one of the earliest deep learning methods. Deep learning gained additional interest in today's applications due to its capability to handle complex and high dimensional problems. In this study, multi-layer GMDH networks are used to perform uncertainty quantification (UQ) and sensitivity analysis (SA) of nuclear reactor simulations. GMDH is utilized as a surrogate/metamodel to replace high fidelity computer models with cheap-to-evaluate surrogate models, which facilitate UQ and SA tasks (e.g. variance decomposition, uncertainty propagation, etc.). GMDH performance is validated through two UQ applications in reactor simulations: (1) low dimensional input space (two-phase flow in a reactor channel), and (2) high dimensional space (8-group homogenized cross-sections). In both applications, GMDH networks show very good performance with small mean absolute and squared errors as well as high accuracy in capturing the target variance. GMDH is utilized afterward to perform UQ tasks such as variance decomposition through Sobol indices, and GMDH-based uncertainty propagation with large number of samples. GMDH performance is also compared to other surrogates including Gaussian processes and polynomial chaos expansions. The comparison shows that GMDH has competitive performance with the other methods for the low dimensional problem, and reliable performance for the high dimensional problem.

• 천문학회보
• /
• 제35권1호
• /
• pp.26.1-26.1
• /
• 2010

In this talk we outline the current understanding of solar flares, mainly focusing on magnetohydrodynamic (MHD) processes. A flare causes plasma heating, mass ejection, and particle acceleration which generates high-energy particles. The key physical processes producing a flare are: the emergence of magnetic field from the solar interior to the solar atmosphere (flux emergence), formation of current-concentrated areas (current sheets) in the corona, and magnetic reconnection proceeding in a current sheet to cause shock heating, mass ejection, and particle acceleration. A flare starts with the dissipation of electric currents in the corona, followed by various dynamic processes that affect lower atmosphere such as the chromosphere and photosphere. In order to understand the physical mechanism for producing a flare, theoretical modeling has been develops, where numerical simulation is a strong tool in that it can reproduce the time-dependent, nonlinear evolution of a flare. In this talk we review various models of a flare proposed so far, explaining key features of individual models. We introduce the general properties of flares by referring observational results, then discuss the processes of energy build-up, release, and transport, all of which are responsible for a flare. We will come to a concluding viewpoint that flares are the manifestation of the recovering and ejecting processes of a global magnetic flux tube in the solar atmosphere, which has been disrupted via interaction with convective plasma while rising through the convection zone.

• Journal of Astronomy and Space Sciences
• /
• 제6권1호
• /
• pp.1-15
• /
• 1989

Magnetospheic substorm in the magnetotail region is studied numerically by means of a three dimensional MHD code. The analytic solution for the quiet magnetotail is employed as an initial configuration. The localized solar wind is modeled to enter the simulation domain through the boundaries located in the magnetotail lobe region. As a result of the interaction between the solar wind and the magnetosphere, the magnetic field lines are stretched, and the plasma sheet becomes thinner and thinner. When the current driven resistivity is generated, magnetic reconnection is triggered by this resistivity. The resulting plasma jetting is found to be super-magnetosonic. Although the plasmoid formation and its tailward motion is not quite clear as in the two dimensional simulation, which is mainly because of the numerical model chosen for the present simulation, the rarification of the plasmas near the x-point is observed. Field aligned currents are observed in the late expansive stage of the magnetospheric substorm. These field aligned currents flow from the tail toward the ionosphere on the dawn side from the ionosphere to ward the tail on the dusk side, namely in the same sense of the region 1 current. As the field aligned currents develop, it is found that the cross tail current in the earth side midnight section of the magnetic x-point is reduced.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1998년도 제13차 학술회의논문집
• /
• pp.195-200
• /
• 1998

Since the neural network was introduced, significant progress has been made on data handling and learning algorithms. Currently, the most popular learning algorithm in neural network training is feed forward error back-propagation (FFEBP) algorithm. Aside from the success of the FFEBP algorithm, polynomial neural networks (PNN) learning has been proposed as a new learning method. The PNN learning is a self-organizing process designed to determine an appropriate set of Ivakhnenko polynomials that allow the activation of many neurons to achieve a desired state of activation that mimics a given set of sampled patterns. These neurons are interconnected in such a way that the knowledge is stored in Ivakhnenko coefficients. In this paper, the PNN model has been developed using the plasma enhanced chemical vapor deposition (PECVD) experimental data. To characterize the PECVD process using PNN, SiO$_2$films deposited under varying conditions were analyzed using fractional factorial experimental design with three center points. Parameters varied in these experiments included substrate temperature, pressure, RF power, silane flow rate and nitrous oxide flow rate. Approximately five microns of SiO$_2$were deposited on (100) silicon wafers in a Plasma-Therm 700 series PECVD system at 13.56 MHz.