Seunghyok Kim

On The Multivariable Robust Control of a Boiler-Turbine System

Abstract In this paper, a multivaribale robust controller for the boiler-turbine system is designed by using the LQG/LTR method. From the known nonlinear dynamic model, a linearized model is obtained with the saturations at both control inputs and their rates and the modeling error is analyzed at various operating points. The controller is designed using the LQG/LTR method to reject the input and output disturbances and to follow reference inputs under modeling errors. A dynamics augmentation method in the LQG/LTR method is suggested. The gpod performance of the designed controller is shown by simulations in various conditions.

Simulation of commercial dimethyl ether production plant

The process of Dimethyl ether(DME) production consists of the four parts which are syngas synthesis from natural gas, absorbing CO2 from syngas, DME synthesis reactor and DME separation/purification. KOGAS has developed a process in which syngas is produced from natural gas and converted to DME using a single reactor[1]. For the construction of commercial scale DME plant, the modeling of one-step DME synthesis reactor is required prior to beginning the construction. Since then, the simulation of DME production process should represent actual operation data of pilot or demo scale plant. The simulations of reactor had been conducted using a one-dimensional steady-state model of a shell-and-tube type fixed-bed reactor[2]. Using the result of a reactor analysis, we have conducted simulations of all processes using steady-state models in Aspen Plus®. The simulation of process in this paper reflects the result of a reactor simulation and the real operation data of demo scale DME plant. And the simulation results are satisfied with the requirements for the basic design and engineering of commercial DME plant construction.

Optimization of sustainable energy planning with consideration of uncertainties in learning rates and external cost factors

Abstract Traditionally, energy planning has focused on meeting the domestic demand with low total costs. However, issues and concerns about global warming have made the problem of energy planning more complex. Since conventional energy supplies cause environmental problems like greenhouse effect, most governments are now trying to increase investment in developing sustainable energy systems (e.g. hydro, wind, solar, and/or biomass energy). Unfortunately, the expansion of promising sustainable energy technologies in the market or research area is not obvious. The production costs of these are uncertain and much higher than those of fossil fuels. To overcome these problems, the concept of learning rate is adopted in this study. And the learning rate helps evaluate the costs of installing and operating sustainable energy systems. In this paper, economic evaluation of conventional power plants and prospective sustainable energy systems with respect to production cost is studied. The production cost includes capital, fixed, variable and external costs. For case study, the data reported by Korean government is used. Using the proposed methodology, authorities and companies will be able to decide the optimal investment and installation plans for sustainable energy supplies.

Principal Component Analysis Based Support Vector Machine for the End Point Detection of the Metal Etch Process

Abstract An endpoint detection using the algorithm of principal component analysis based support vector machine was developed for the plasma etching process. Because many endpoint detection techniques use a few manually selected wavelengths, noise render them ineffective and it is hard to select the important wavelengths. So the principal component algorithm with the whole wavelengths has been developed for the more effective monitoring of end point. And the support vector regression was followed for the realtime end point detection with reduced wavelengths to save the processing time. This approach was applied and demonstrated for a metal etching process of Al and 0.5% Cu on the oxide stack with inductively coupled BCl 2 /Cl 2 plasma.

Evaluation of coal-based dimethyl ether production system using life cycle assessment in South Korea

Abstract Dimethyl ether is one of the sustainable energy sources because it is clean and easy to handle. DME can be produced from various materials such as coal, natural gas (NG), and biomass. Actually, NG-based DME plant is in operation in Korea. Many studies have also investigated NG-based processes and reactor design needed for use of dimethyl ether. Unlike the previous studies, this study focuses on DME production system using coal as a raw material. Coal is drawing attention as the upcoming energy source due to the depletion of oil in Korea. Whereas Korea depends entirely on imports for crude oil, coal can be supplied by domestic production. To decide coal as a raw material, we need to assess the competitiveness of coal. A life cycle assessment of coal-based DME production system was performed to evaluate environmental impact and energy consumption. The assessment of coal-based DME production system with respect to the environment and energy use is essential for sustainable DME production in Korea. The aim of this study was to analyze the amount of CO2 emission and energy consumption of coal-based DME production. The life cycle of DME production includes the mining of coal, pretreatment of the mined coal, transportation to the DME production plant, and DME conversion. Related data and the assumptions pertinent to South Korea are introduced and adopted as a case study.

Systematic approach towards inspecting equipment defects in chemical plants

Abstract Today, there are so many chemical plants in the world. The more chemical plants are established, the more it is important to maintain them properly. The inspection and maintenance of risk equipments in chemical plants are of fundamental importance in process safety system; however, strategic methods in maintaining equipments like vessels, pipes and valves are applied on few plants only. This paper presents the systematic approach towards inspecting equipment (pipe lines centrally) defects in chemical plants. It is accomplished by analyzing the inspection and maintenance records in specific plants (specific data), as well as the offered standard database (generic data) first. Then, equipment that is frequently defected can be distinguished. By using the Dow Fire and Explosion Index (DF&EI), risk index of the process units is generated. In addition, risk priority of equipment is generated by applying the damage mechanism standards. As a result, the inspection priority based on risk can be achieved that is based on semi-quantitative method.

Evolutionary Improvement Algorithm with Statistical Learning Method for Process Real-Time Optimization

Abstract This study proposes an effective framework for process real-time optimization and data-driven modeling method. The proposed RTO framework with evolutionary improvement algorithm does not wait for the steady-state and it corrects the set-point continuously through the similar way which genetic algorithm exploit to find optimal points. It can deal with higher frequency disturbances and is less influenced by control system performance. Moreover, it is able to address the convergence to suboptimal. Also, this study proposes statistical learning model (modified Support Vector Machine) that is used in RTO framework. It is able to handle highly-nonlinearity and carry out parameter tuning easily. The performance of proposed method was successfully illustrated by means of RTO example.

ON THE MULTIVARIABLE ROBUST CONTROL OF A BOILER-TURBINE SYSTEM

In this paper, a multivaribale robust controller for the boiler-turbine system is designed by using the LQG/LTR method. From the known nonlinear dynamic model, a linearized model is obtained with the saturations at both control inputs and their rates and the modeling error is analyzed at various operating points. The controller is designed using the LQG/LTR method to reject the input and output disturbances and to follow reference inputs under modeling errors. A dynamics augmentation method in the LQG/LTR method is suggested. The good performance of the designed controller is shown by simulations in various conditions.