Title: New approaches for experiments with mixtures
Authors: Stefanie Biedermann - University of Southampton (United Kingdom) [presenting]
Steven Gilmour - KCL (United Kingdom)
Rana Khashab - Umm Al-Qura University (Saudi Arabia)
Abstract: Experiments involving mixtures are conducted in a variety of areas, for example in food processing or in chemical research. The experimental region is constrained naturally, as the proportions of all ingredients have to sum to one. Additional constraints may arise when there are bounds on the proportions, for example a cake must contain a minimum percentage of flour to have the right texture and flavour. We propose a new - parsimonious but flexible - class of non-linear models, based on fractional polynomials, to fit the data from constrained mixture experiments. We will motivate this modelling approach, and will use a number of historical data sets to compare these models with various other models suggested in the literature. We will then present some optimal designs for these models, and will further discuss some general issues related to designing experiments for mixtures.