Analysis of the variability of food texture properties: Application to the fracturability of dry pet food. Issue 6 (25th May 2017)

• Record Type: Journal Article
• Authors: Gouriou, Benjamin; Bressolette, Philippe; Vial, Christophe
• Is Part Of: Journal of texture studies. Volume 48:Issue 6(2017)
• Page Start: 586
• Page End: 596
• Publication Date: 2017-05-25
• Subjects: pet food -- screening method -- stochastic finite elements -- texture analysis -- variability; Food texture -- Periodicals; Food -- Composition -- Periodicals; 664.02
• DOI: 10.1111/jtxs.12269
• Languages: English
• ISSNs: 0022-4901

Abstract

Abstract: An original stochastic finite element method (SFEM) is proposed for the texture analysis of food products for which samples cannot be standardized. SFEM is able to distinguish shape and texture variability. As an illustration, the methodology is applied to dry cat food using compression testing. First, a deterministic damage elastic model depending on six mechanical and three size parameters is shown to fit adequately experimental data. Then, Morris screening method is applied to FEM data: this highlights that the variability of compression tests is significantly affected by two mechanical and two size parameters. Finally, the nonlinear variability of each control variable is estimated from Sobol indices derived from a time‐efficient stochastic collocation method. These highlight that the contribution of size parameters to test variability is about 20%. Therefore, a robust estimation of the probability density function of texture properties can be obtained for improved quality control, which outperforms current methods. Practical applications: Food texture is affected by their material property and geometrical structure. Material properties are difficult to estimate when the size and shape of the samples cannot be standardized, for example, for snack or pet food. In addition, the intrinsic variability of material properties due to the recipe and the process cannot be neglected. But this cannot be easily distinguished from the variability of geometrical parameters using conventional methods. An innovative approach using stochastic finite element method is developed and is able to exhibit the major parameters influencing the variability of the response to mechanical testing. This methodology may be helpful to improve acceptance sampling control procedures, for example, by combining texture and size measurements when necessary, or by better determining the most robust response variables for decision‐making.