Mixtures

Table of contents

  1. Mixtures Design
  2. References
  3. Version History

Mixtures Design

Mixture design (experiment) is a special class of response surface method (RSM). These experiments focus on understanding how various proportions of components influence the overall characteristics of a mixture. A critical aspect of mixture experiments is that the factors being tested are the proportions of each component within the mixture. There’s an inherent constraint in these experiments: the sum of the proportions of all components must always equal 1. This constraint introduces unique challenges in both the design and analysis phases of the experiments.

To effectively address these challenges, two popular designs are often employed: the simplex lattice design and the simplex centroid design. The simplex lattice design offers a comprehensive approach by testing at all possible combinations of component proportions, including those at the vertices, edges, and faces of the simplex. This design allows for a detailed exploration of the interactions between components across the entire range of possible formulations.

In contrast, the simplex centroid design focuses on the centroid of each subspace of the simplex, including the overall centroid. This method is particularly beneficial for initial studies where a full-scale experimental approach may be impractical. It provides a more streamlined set of experiments that can still yield valuable insights into the main effects and some interactions among the mixture components.

Both the simplex lattice and simplex centroid designs are instrumental in overcoming the challenges posed by the fixed sum constraint in mixture experiments. They enable researchers to systematically explore and optimize different formulations, making them invaluable tools in the development of new products and processes where mixture proportions are critical.

Use the Mixtures function by browsing in the top ribbon:

DOE \(\rightarrow\) Mixtures \(\rightarrow\) Mixtures Design

Input

Input data for Mixtures design is not required.

Configuration

Number of Blocks Select manually the Number of Blocks. The lowest value that can get is 1.
Random Standard order You can tick/select the box if randomness is required in the output list of experiments.
Time-based RNG Seed If randomness is selected, then you have the option to also tick/select the box to set the random generated number seed based on time.
RNG Seed Select manually the random generated number seed if required.
Select Design Select Design offers two options either Simplex Centroid or Simplex Lattice.
Number of components Select the Number of components (factors) from the drop-down list.
Degree of Lattice If Simplex Lattice design is chosen, then select the Degree of Lattice from the drop-down list.
Augment the Design with Axial Points Select/tick the box to Augment the Design with Axial Points if desired.
Augment the Design with Center Points If Simplex Lattice design is chosen, then there is also the option to select/tick to Augment the Design with Center Points.
Component Drag the horizontal value bars for each component to specify the min and max (composition) value for each component.

Output

A list of experiments (combinations) is generated in the output datasheet along with the Block number and the Standard order based on the Mixtures design method selected.

Example

Input

No input data is required in the input datasheet.

Configuration
  1. Select DOE\(\rightarrow\) Mixtures \(\rightarrow\) Mixtures Design.
  2. Select the Number of Blocks [1]. The lowest value that can be set is 1.
  3. Select/tick if required the Random Standard order [2] to imply randomness in the output list of experiments.
  4. If randomness is selected either select/tick to generate the number seed for randomness based on time [3] or by manually setting a value [4].
  5. Select Design to be either Simplex Centroid or Simplex Lattice [5].
  6. Select Number of components in the Mixture from the drop-down list [6].
  7. If Simplex Lattice design method is selected, then select the Degree of Lattice from the drop-down list [7].
  8. Select/tick to Augment the Design with Axial Points [8].
  9. If Simplex Lattice design method is selected, then select/tick to Augment the Design with Center Points [9].
  10. Drag the horizonal value bars for each component [10] and [11] to specify the min and max (composition) value for each component.
  11. Click on the Execute button [12] to perform Mixtures design method.
Mixtures
Output

The output list of experiments is generated for the two components outlining a list of 4 experiments based on the Mixtures design. Block number is presented in Col1 as shown below and the standard order is provided in Col2.

Mixtures output

References

  1. Box, G.E. and Behnken, D.W., Some new three level designs for the study of quantitative variables. Technometrics, 1960. 2(4): p. 455-475. doi.org/10.1080/00401706.1960.10489912.

  2. Box, G.E. and Wilson, K.B., On the experimental attainment of optimum conditions, in Breakthroughs in statistics: methodology and distribution. 1992, Springer. p. 270-310. doi.org/10.1007/978-1-4612-4380-9_23.

  3. Montgomery, Douglas C. Design and analysis of experiments. John wiley & sons, 2017.

  4. Speed, Terry. Statistics for Experimenters: Design, Innovation, and Discovery. 2006.

  5. Wu, CF Jeff, and Michael S. Hamada. Experiments: planning, analysis, and optimization. John Wiley & Sons, 2011.

Version History

Introduced in Isalos Analytics Platform v0.2.4

Instructions last updated on January 2025