Sequential experimentation approach for robust design

Ríos, Armando J.; Guerrero, Guadalupe

doi:10.1002/qre.2332

Cited by 3 publications

(2 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some related studies on sequential experimentation include Rios 41 in which the R3 algorithm was used to separate MEs from 2FIs in resolution III fractions and Misra 42 in which a technique called Quarterfold was used to separate 2FIs from each other in resolution IV fractions. In addition, Rios 43 presented a method to eliminate multicollinearity and Rios 44 presented a sequential experimentation approach for robust designs. A direct comparison with these studies is presented in Table 17.…”

Section: Comparison Of the Proposed Methods Against Other Techniquesmentioning

confidence: 99%

A sequential experimentation method to separate interaction effects from block effects

Ríos

Murillo

Pantoja

et al. 2020

Engineering Reports

Self Cite

View full text Add to dashboard Cite

Blocking is a basic experimentation principle that separates the variability caused by noise factors. Unless the experiment is replicated, blocking produces loss of information, particularly two‐factor interactions (2FIs) are commonly lost to blocks. Additionally, the ANOVA table does not show to what extent each blocked noise factor affects the response variable. Individual contributing percentages for noise factors can be useful to make process improvements and to understand which noise factors are most influential. This research proposes a sequential experimentation method to separate 2FIs from blocks and assign contributing percentages to each blocked noise factor. The method is evaluated and compared to foldover, semifold, and D‐optimal augmentation.

show abstract

Section: Comparison Of the Proposed Methods Against Other Techniquesmentioning

confidence: 99%

A sequential experimentation method to separate interaction effects from block effects

Ríos

Murillo

Pantoja

et al. 2020

Engineering Reports

Self Cite

View full text Add to dashboard Cite

show abstract

“…(iii) Type II error is the number of significant terms missing in the model [36]. % Type II measure the percentage of times that the models showed significant missing terms during the simulations:…”

Section: Run Factorsmentioning

confidence: 99%

Quantitative Analysis of the Balance Property in Factorial Experimental Designs 24 to 28

et al. 2022

View full text Add to dashboard Cite

Experimental designs are built by using orthogonal balanced matrices. Balance is a desirable property that allows for the correct estimation of factorial effects and prevents the identity column from aliasing with factorial effects. Although the balance property is well known by most researchers, the adverse effects caused by the lack or balance have not been extensively studied or quantified. This research proposes to quantify the effect of the lack of balance on model term estimation errors: type I error, type II error, and type I and II error as well as R2, R2adj, and R2pred statistics under four balance conditions and four noise conditions. The designs considered in this research include 24–28 factorial experiments. An algorithm was developed to unbalance these matrices while maintaining orthogonality for main effects, and the general balance metric was used to determine four balance levels. True models were generated, and a MATLAB program was developed; then a Monte Carlo simulation process was carried out. For each true model, 50,000 replications were performed, and percentages for model estimation errors and average values for statistics of interest were computed.

show abstract

Diseños ortogonales de Taguchi fraccionados

Naranjo-Palacios

Ríos-Lira

Pantoja-Pacheco

et al. 2020

iit

View full text Add to dashboard Cite

El diseño de experimentos es una herramienta utilizada para descubrir como entran en juego distintas variables de un proceso en la obtención de un producto. Existen dos enfoques principales para realizar experimentación, el enfoque clásico y el enfoque de Taguchi. Los diseños de Taguchi son diseños ortogonales que se especializan en estimar efectos principales e interacciones de control por ruido, dejando en segundo plano las interacciones de control por control. Los arreglos ortogonales de Taguchi fueron diseñados de tal manera que un arreglo especifico puede ser utilizado para diferentes números de factores, por ejemplo, el L32 se utiliza cuando existen de 16 a 31 factores y requiere de 32 experimentos. Cuando el número de columnas disponibles excede al número de factores que se desea investigar, las columnas sobrantes se utilizan comúnmente para estimar interacciones. Sin embargo, en casos en que el investigador esta solo interesado en los efectos principales, correr el arreglo completo podría ser algo innecesario y costoso. La presente investigación tiene como objetivo fraccionar los arreglos ortogonales de Taguchi L8, L12, L16 y L32 de tal forma que la fracción generada sirva únicamente para estimar efectos principales y las corridas restantes se agreguen solo en caso de ser requeridas. El método propuesto se basa en búsqueda exhaustiva y utiliza como criterios de selección la D-optimalidad, los factores de inflación de varianza (FIV) y el índice de balance general (IBG). Solo arreglos ortogonales de Taguchi de dos niveles fueron considerados para esta investigación. Los resultados de la investigación se traducen en ahorros significativos de recursos, reducción del tiempo de experimentación y del numero de corridas.

show abstract

Sequential experimentation approach for robust design

Cited by 3 publications

References 16 publications

A sequential experimentation method to separate interaction effects from block effects

A sequential experimentation method to separate interaction effects from block effects

Quantitative Analysis of the Balance Property in Factorial Experimental Designs 24 to 28

Diseños ortogonales de Taguchi fraccionados

Contact Info

Product

Resources

About