Premium
D ‐Optimal Experimental Designs for Uniaxial Expression
Author(s) -
MunsonMcGee Stuart H.
Publication year - 2014
Publication title -
journal of food process engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.507
H-Index - 45
eISSN - 1745-4530
pISSN - 0145-8876
DOI - 10.1111/jfpe.12080
Subject(s) - expression (computer science) , variance (accounting) , design of experiments , optimal design , sine , computer science , process (computing) , series (stratigraphy) , exact solutions in general relativity , algorithm , mathematics , mathematical optimization , statistics , mathematical analysis , paleontology , geometry , accounting , business , biology , programming language , operating system
D‐optimal designs are developed for uniaxial expression of oil from seeds, nuts and other materials. Both simply supported pseudo‐exact and pseudo‐continuous designs are determined for three variations on the expression model: instantaneous pressure application, ramped pressure application and sine wave pressure application. The results show that the designs are dependent on the magnitude of the applied pressure and, to a much lesser extent, the rate at which it is applied and the manner in which it is applied. For most cases, the maximum standardized variance for the simply supported pseudo‐exact designs was 1.005 or less. The pseudo‐continuous designs were only significantly better when the maximum standardized variance of the simply supported pseudo‐exact designs was greater than 1.02. Practical Applications This work presents a series of new experimental designs for the expression modeling process that can aid researchers in developing more efficient designs. The designs developed in this work integrate two separate existing fields of study: D ‐optimal design theory and theoretical expression modeling. Furthermore, the results show under what combinations of the model parameters the simply supported, pseudo‐exact designs can be used instead of the more complex pseudo‐continuous designs. For experiments that do not use the model parameters considered herein, the complete development of the numerical approach provides the framework for others to develop the appropriate designs to satisfy their needs.