• Biotechnology and Bioprocess Engineering:BBE
• /
• v.10 no.6
• /
• pp.471-481
• /
• 2005

Heavy crudes (bitumen) are extremely viscous and contain high concentrations of asphaltene, resins, nitrogen and sulfur containing heteroaromatics and several metals, particularly nickel and vanadium. These properties of heavy crude oil present serious operational problems in heavy oil production and downstream processing. There are vast deposits of heavy crude oils in many parts of the world. In fact, these reserves are estimated at more than seven times the known remaining reserves of conventional crude oils. It has been proven that reserves of conventional crude oil are being depleted, thus there is a growing interest in the utilization of these vast resources of unconventional oils to produce refined fuels and petrochemicals by upgrading. Presently, the methods used for reducing viscosity and upgradation is cost intensive, less selective and environmentally reactive. Biological processing of heavy crudes may provide an ecofriendly alternative or complementary process with less severe process conditions and higher selectivity to specific reactions to upgrade heavy crude oil. This review describes the prospects and strengths of biological processes for upgrading of heavy crude oil.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.19 no.10
• /
• pp.2690-2698
• /
• 1995

Fuel atomization and mixture formation in an gasoline engine has influence on the engine performance and pollutant emission. The throttle valve installed in an intake system plays a greater role in control of mixture quantity in accordance with engine drive condition. In this study, the characteristics of secondary atomization developed at the downstream of the valves were observed using an image processing method. Two major kinds of valves, solid and perforated ones, are chosen in order to compare the valve performance with the experimental parameters of air flow rate, valve opening angle, and valve shapes. For the perforated valve, we can obtain the relatively small sized droplets, and nearly uniformed and dense distributed sprays with low loss coefficient than for the solid valve.

• Proceedings of the KIEE Conference
• /
• 1994.07b
• /
• pp.1609-1611
• /
• 1994

In ECR and helicon reactors for plasma processing, a high density plasma is generated in a source region which is connected to a diffusion region where the processing takes place. Large density and potential gradients can develop at the orifice of the source which drive ion currents into the diffusion region. The average ion velocity may become the order of the sound velocity. Measurements of the ion saturation current to a Langmuir probe are used as a standard method of determining the plasma density in laboratory discharges. However, the analysis becomes difficult in a steaming plasma. We have used the HAMLET plasma simulator to simulate the ion flow to a large langmuir probe in an ECR plasma. The collection surface was aligned with the Held upstream, normal to the field, and downstream. ion trajectories through the electric and magnetic fields were calculated including ion-neutral collisions. We examines the ratio of ion current density to plasma density as a function of magnetic field and pressure.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2001.04a
• /
• pp.511-514
• /
• 2001

본 논문에서는 Ad Hoc 네트워크에서의 라우팅 알고리즘인 TORA 의 라우팅 유지관리 방법에서 발생하는 비효율적인 컨트롤 패킷들을 줄이는 방법을 제안한다. Adv-TORA에서는 DN(Downstream)이 소멸되는 노드에서 STP(Stop) 패킷을 주변 노드에 발송하여 끊어진 경로로 데이터 패킷이 발송되는것을 사전에 방지한다. 또, DN 소멸 노드의 주위에서 그래프의 재구성 작업이 일어났던 TORA 와는 상이하게, 단순히 지역적 라우트 경로를 초기화시킴으로서 TORA 의 컨트롤 패킷 부담을 줄인다. 또한, 통신에 참여하지 않는 노드들의 불필요한 라우트 정보 유지를 최소화하고, 라우트 요구 발생시에만 라우트 재구성 작업을 수행한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.678-680
• /
• 2023

본 논문은 정확하면서도 효율적인 한국어 문장 분류 기법에 대해서 논의한다. 최근 자연어처리 분야에서 사전 학습된 언어 모델(Pre-trained Language Models, PLM)은 미세조정(fine-tuning)을 통해 문장 분류 하위 작업(downstream task)에서 성공적인 결과를 보여주고 있다. 하지만, 이러한 미세조정은 하위 작업이 바뀔 때마다 사전 학습된 언어 모델의 전체 매개변수(model parameters)를 학습해야 한다는 단점을 갖고 있다. 본 논문에서는 이러한 문제를 해결할 수 있도록 도메인 적응기(domain adapter)를 활용한 한국어 문장 분류 프레임워크인 DAKS(Domain Adaptation-based Korean Sentence classification framework)를 제안한다. 해당 프레임워크는 학습되는 매개변수의 규모를 크게 줄임으로써 효율적인 성능을 보였다. 또한 문장 분류를 위한 특징(feature)으로써 한국어 사전학습 모델(KLUE-RoBERTa)의 다양한 은닉 계층 별 은닉 상태(hidden states)를 활용하였을 때 결과를 비교 분석하고 가장 적합한 은닉 계층을 제시한다.

• The Journal of Fisheries Business Administration
• /
• v.52 no.4
• /
• pp.45-63
• /
• 2021

The purpose of this study is to analyze the structure, status and economic ripple effects of the fisheries processing industry in Korea using interindustry analysis. Five input-output tables published over the past twenty years have been reclassified with a focus on the fisheries processing sector. Through these multi-period tables, we analyzed changes in the inducing effects in production, value added and employment as well as the backward-forward linkage effects. As a result of the analysis, it was found that the industrial scale of the fisheries processing industry is very small compared to other food manufacturing industries. The backward linkage effect of the fisheries processing industry was greater than that of other industries, but the forward linkage effect was rather low. This means that the fisheries processing industry can be greatly affected by industrial depression of the downstream industries such as fishery and aquaculture. Production and employment-inducing effects of the fisheries processing industry have shown a decreasing trend in recent years. This reflects the reality that intermediate inputs are gradually being replaced by imports from domestic production due to the expansion of market opening and the depletion of fishery resource. In the future, it is necessary to prepare a strategy to increase the value-added productivity of the fisheries processing sector and foster it as an export industry.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.43 no.11 s.353
• /
• pp.149-158
• /
• 2006

We proposed a method for improving a performance of TCP downstream between a desktop PC as a fixed host and a PDA as a mobile host in a wired and wireless network based on IEEE 802.11x wireless LAN. With data transmission between these heterogeneous terminals a receiving time during downstream is slower than that during upstream by 20% at maximum. The reason is that their congestion window size will be oscillated due to a significantly lower packet processing rate at receiver compared to a packet sending rate at sender. Thus it will cause to increase the number of control packets to negotiate their window size. To mitigate these allergies, we proposed two distinct methods. First, by increasing a buffer size of a PDA at application layer an internal processing speed of a socket receive buffer of TCP becomes faster and then the window size is more stable. However, a file access time in a PDA is kept nearly constant as the buffer size increases. With the buffer size of 32,768bytes the receiving time is faster by 32% than with that of 512bytes. Second, a delay between packets to be transmitted at sender should be given. With an inter-packet delay of 5ms at sender a resulting receiving time is faster by 7% than without such a delay.

• Korea-Australia Rheology Journal
• /
• v.16 no.1
• /
• pp.27-33
• /
• 2004

Application of gas assisted injection molding has been expanded during last two decades because of many advantages such as design flexibility, dimensional stability, reduction of machine tonnages, and so on. However, the surface defects including hesitation mark and gloss difference are observed for thick parts. Difficulties in lay-out of the gas channel and processing condition are another disadvantages. Liquid gas assisted injection molding(LGAIM), in which a liquid with a boiling point lower than the temperature of the polymer melt is injected into the melt stream, and travels with the melt into the mold where it vaporizes and pushes the melt downstream and against the cavity walls to create hollow channels within the part, is a good alternative of the conventional gas assisted injection molding especially in manufacturing simple and very thick parts. Though this is a new frontier of the innovation in the injection molding industry, there is no guideline for the design and processing conditions. In this paper, theoretical analysis has been made to describe the hollow formation dynamics in LGAIM. This model provides an insight into LGAIM process: explains why LGAIM has advantages over conventional gas assisted injection molding, and gives a guideline for the design and processing conditions.

• Korean Journal of Optics and Photonics
• /
• v.9 no.4
• /
• pp.264-269
• /
• 1998

The Measurement of concentration patterns for $C_2$, CH and OH radicals in the premixed propane-air flame could be managed through an image processing technique. This technique was applied to the three kinds of flames on a bunsen burner-mixtures of fuel to be lean, optimum and excessive respectively. The image processing system was complished by treating single wavelength flame images around the eac radical luminescence band, which was obtained by using a set of narrow band pass filters, an image intensifier, CCD and PC. It was possible to observe and predict the reaction zone and the concentration distribution of the radicals, Spatial distribution of each radicals in the raaction zone gave us enough informations to analyze the reaction mechanisms in $C_mH_n$ combustion process. According to this informations, the image of $C_2$ radical exists at front zone, following the images of CH and OH radicals at downstream.

• Journal of Korean Institute of Industrial Engineers
• /
• v.24 no.4
• /
• pp.649-660
• /
• 1998

Steel making in Mini Mill consists of three major processing stages: molten steel making in an electric arc fuenace, slab casting in a continuous caster, and hot rolling in a finishing mill. Each processing stage has its own lot grouping criterion. However, these criteria in three stages are conflicting with each other. Therefore, delveloping on efficient lot grouping algorithm to enhance the overall productivity of the Mini Mill is an extremely difficult task. The algorithm proposed in this paper is divided into three steps hierarchically: change grouping, cast grouping, and roll grouping. An efficient charge grouping heuristic is developed by exploiting the characteristics of the orders, the processing constraints and the requirements for the downstream stages. In order to maximaize the productivity of the continuous casters, each cast must contain as many charges as possible. Based on the constraint satisfaction problem technique, an efficient cast grouping heuristic is developed. Each roll consists of two casts satisfying the constraints for rolling. The roll grouping problem is formulated as a weighted non-bipartite matching problem, and an optimal roll grouping algorithm is developed. The proposed algorithm is programmed with C language and tested on a SUN Workstation with real data obtained from the H steel works. Through the computational experiment, the algorithm is verified to yield quite satisfactory solutions within a few minutes.