• 대한안전경영과학회:학술대회논문집
• /
• 대한안전경영과학회 2011년도 춘계학술대회
• /
• pp.487-493
• /
• 2011

The paper reviews three Split Plot Designs (SPDs) of fixed factors, and those are SPD (RCBD, RCBD), SPD (CRD, RCBD) and SBD (Split Block Design). RCBD (Randomized Complete Block Design) and CRD (Completely Randomized Design) are used to deploy whole plot and sub plot. The models explained in this study are derived from random, crossed and nested models.

• 대한안전경영과학회:학술대회논문집
• /
• 대한안전경영과학회 2009년도 추계학술대회
• /
• pp.649-657
• /
• 2009

The paper presents the measurement split split-plot models for saving the time and cost. The split split-plot designs developed are efficiently used to estimating the gauge R&R(Reproducibility & Repeatability) when the completely randomized design of all factors(such as high pressure and temperature) is expensive and time consuming. The models studied include three split split-plots considering the type of experimental units.

• Communications for Statistical Applications and Methods
• /
• 제12권1호
• /
• pp.61-69
• /
• 2005

In conducting the analysis of variance for the split-plot design using the statistical package SPSS, users including statisticians are faced with difficulties because of no appropriate example in the SPSS applications guide book. In this paper, therefore, we present an analysis of variance procedure for the split-plot design using SPSS syntax window.

• 대한안전경영과학회:학술대회논문집
• /
• 대한안전경영과학회 2012년 춘계학술대회
• /
• pp.277-292
• /
• 2012

The research presents an application of Balanced ANOVA (BANOVA) by utilizing randomized orders for various Split-Plot Designs (SPDs) which include two cell designs, split-plot with one-way HTC (Hard to Control) factor, split-plot with two-way HTC factor, split-split-plot design and nested design. In addition, four MINITAB examples of 2-level split-plot designs based on the number of blocks and the type of whole-plots are presented for practitioners to obtain comprehensive understanding. Furthermore, the geometrical interrelated properties among three typical Designs of Experiments (DOE), such as Factorial Design (FD), Response Surface Design (RSD), and Mixture Design (MD) are discussed in this paper.

• 산업기술연구
• /
• 제31권A호
• /
• pp.27-31
• /
• 2011

In many industrial experiments, some practical constraints often force factors in an experiment to be much harder to change than others. Such an experiment involves randomization restrictions and it can be thought of as split-plot experiment. This paper investigates the path of steepest ascent/descent within a split-plot structure. A method is proposed for calculating the coordinates along the path.

• 대한안전경영과학회:학술대회논문집
• /
• 대한안전경영과학회 2011년도 춘계학술대회
• /
• pp.481-486
• /
• 2011

The research reviews three experimental designs which include Split Plot Design (SPD), Crossover Design (CD) and Replicated Design (RD) by using Latin Square Design (LSD). SPD (CRD, LSD) and SPD (LSD, RCBD) that are derived from (S:A)${\times}B{\times}C{\times}D$ and $A{\times}B{\times}C{\times}D$. In addition, (S:A)${\times}B{\times}C$, (S:A)${\times}C{\times}D$ and (S:A)${\times}B{\times}C{\times}D$ can be used to generate various LSD and CD models. Finally, Replicated LSDs are considered to increase the power of detectability.

• Journal of the Korean Data and Information Science Society
• /
• 제28권1호
• /
• pp.143-152
• /
• 2017

블럭내의 완전랜덤화 제약은 하나의 블럭이 여러 실험구로 분할되는 분할요인모형으로 해결할 수 있다. 본 연구는 $3{\times}3$ 분할요인모형에서 두 주요인 및 하나의 블럭이 모두 고정일 경우에는, 실제로 존재하는 효과크기가 작을수록 혹은 검정대상의 요인효과 크기보다 검정대상 이외의 효과들의 크기가 상대적으로 작을수록 주구요인효과 및 세구요인효과 검정을 위한 순위변환 통계량의 검정력은 기존의 모수적 통계량의 검정력보다 뛰어남을 알 수 있다. 또한 모집단 모형의 오차항이 지수분포 및 이중지수분포일 때 효과크기 및 효과구성유형에 상관없이 거의 모든 상황하에서 순위변환 통계량의 검정력이 모수적 통계량의 검정력보다 상대적으로 높은 우위를 보이며, 정규분포 및 균일분포하에서는 상당히 유사한 수준을 나타낸다. 한편 두 주요인은 고정이나 하나의 블럭이 랜덤일 경우에는, 두 주요인 및 블럭이 모두 고정일 경우보다 모수적 통계량 및 순위변환 통계량의 검정력은 각각 낮은 수준을 보인다. 특히 주구요인효과 검정보다 세구요인효과 검정을 위한 모수적 통계량 및 순위변환 통계량의 검정력이 다소 낮은 수준임을 보이지만, 순위변환 통계량의 검정력은 모수적 통계량의 검정력에 비하여 높은 상대적 검정력 우위를 나타낸다.

• 산업경영시스템학회지
• /
• 제32권4호
• /
• pp.223-227
• /
• 2009

The study presents two measurement split-plot models with a restricted randomization to save cost and time. Split-plot models are used to handle HTCM (Hard To Control Measurement) factors such as high temperature and long-time catalyst control. The models developed are represented by the processes for estimating the measurement precisions, that is, gauge R&R. The study also introduces three-step procedures to indentify resolution, improve R&R reduction, and evaluate the precision effect.

• 품질경영학회지
• /
• 제46권3호
• /
• pp.651-658
• /
• 2018

Purpose: The algorithm described in Lim(1998) is available to determine the sample size directly given specified significance level, power and signal-to-noise ratio. We research on the efficient determination of the sample size by visual methods. Methods: We propose three graphs for investigating the mutual relationship between the sample size r, power $1-{\beta}$ and the detectable signal-to-noise ratio ${\Delta}$. First graph shows the relationship between ${\Delta}$ and $1-{\beta}$ for the given r and it can be checked whether the power is sufficient enough. Second graph shows the relationship between r and ${\Delta}$ for the given power $1-{\beta}$. Third graph shows the relationship between r and $1-{\beta}$ for the given ${\Delta}$. It can be checked that which effects are sensitive to the efficient sample size by investigating those graphs. Results: In factorial design, randomized block design and the split plot design how to determine the sample size directly given specified significance level, power and signal-to-noise ratio is programmed by using R. A experiment to study the split plot design in Hicks(1982) is used as an example. We compare the sample sizes calculated by randomized block design with those by split plot design. By using graphs, we can check the possibility of reducing the sample size efficiently. Conclusion: The proposed visual methods can help an engineer to make a proper plan to reduce the sample size.

• 대한안전경영과학회지
• /
• 제10권1호
• /
• pp.191-195
• /
• 2008

The present study contributes reviewing and suggesting various models for measurement system analysis (MSA). Measurement errors consist of accuracy, linearity, stability, part precision, repeatability and reproducibility (R&R). First, the major content presents split-plot design, and the combination method of crossed and nested design for obtaining gage R&R. Second, we propose $\bar{x}-s$ variable control chart for calculating the gage R&R and number of distinct category. Lastly, investigating the determination of gage performance curve which establishes the control specification propagating calibration uncertainties and measurement errors is described.