Optimized Pet Products Through Process DoE

Background

Design of Experiments (DoE) is the statistically balanced design of conditions or settings hypothesized to impact or affect the variation of or characteristic of a product. It is intended to model the relationships between the independent process variables (factors) over various levels, and the effect on measured product quality attributes or process variability. This model will help achieve the target product, such as appearance, texture, digestibility, and robust process (minimized surging and maximized control) by statistically predicting this optimized outcome.

Pet foods and treats can be optimized or developed by utilizing DoE. Many companies may be missing out on product innovation opportunities by only focusing on ingredient changes. Even by keeping the formula the same, process changes, alone, can open many new product textures and appearances. 

Case Study

There are many dog dental treats on the market and many of which are extruded. Many of these products have unique recipes that can be supported to provide a dental benefit. However, the extrusion operation comprises many process variables that can be adjusted and optimized to create many different product textures that can also be used to demonstrate dental benefits. With DoE, product attributes (surface appearance, size, shape, puncture texture, and digestibility) can be targeted and designed instead of just by chance. The example shown below is what a DoE could be based on through my knowledge of extruded dog dental treats.

Statistical Model Outcome:

COV = Coefficient of Variation, IVD = in vitro digestion, JMP = is the name of a statistical software package

The success of the DoE is linked to the amount of time put into determining the main-effect variables (and their levels), which will significantly affect the key product attributes for your organization. Including both experts and experienced operators, in the design phase, will help ensure that the DoE will encompass the design space of the product. Having too many trials within the experiment may result in no product or data that will negatively impact the model's ability to predict or optimize effectively. When possible, have someone experienced, like BSM Partners, in Six Sigma or DoE around to help with design, execution, and analysis.

Reference

“JMP” statistical software: Design of Experiment

https://www.jmp.com/en_us/software/capabilities/design-of-experiments.html

Example: Six Sigma: “Design of Experiment Study Guide” and course offerings

https://sixsigmastudyguide.com/design-of-experiments-study-guide/

 

Follow us on LinkedIn for the latest updates on all things happening here at BSM Partners.

About the Author

Gregory Daniel is the Director of Process Innovation within the Engineering group at BSM Partners. His technical expertise and a broad range of experiences, from plastic injection molding to over 25 years of pet food and pet treat development, allow him to bring a very unique and different approach to business challenges. In his role, he can combine technical problem-solving, LEAN Six Sigma Black Belt capabilities, and team development skills to drive solutions for any type of problem.