The Challenge
LiveKindly Collective faced significant variability in water uptake during the extrusion process, even when using identical extruders and consistent settings. Water uptake is a crucial quality attribute, directly influencing the texture and juiciness of plant-based meat alternatives. Achieving a consistent water uptake of more than 40% was essential to maintain high product standards.
Despite extensive efforts, variability persisted across production runs, impacting product quality and consistency. To address this, LiveKindly partnered with JADS and EngD candidate Roya Sadeghimehr to systematically analyze and improve the extrusion process using data science.
What Roya Did
Roya developed and implemented a comprehensive data science pipeline to validate extrusion process data, measure water uptake, and analyze how extrusion variables influence the final product. Key activities included:
- Data Validation: Roya ensured the reliability of the data collected through LiveKindly’s SCADA system, identifying and correcting inaccuracies.
- Data Collection: She designed a strategy to measure water uptake during production processes, overcoming data limitations.
- Statistical and Machine Learning Analysis: Roya used methods like Lasso regression and decision trees to identify critical extrusion parameters, such as barrel temperature profiles, cooling die temperature, screw speed, and soy-to-water ratios.
- Actionable Insights: Her analysis revealed that optimizing variables like smoother barrel temperature transitions, higher cooling flows, and specific soy-to-water ratios significantly improved water uptake and overall product quality.
The Result
Roya’s work led to actionable recommendations for optimizing extrusion variables, enhancing process consistency and achieving higher water uptake in plant-based meat. LiveKindly Collective gained a deeper understanding of their extrusion process, enabling them to make data-driven improvements.
Her findings have laid the groundwork for further research and refinement, ensuring more consistent production of high-quality plant-based meat. This successful collaboration underscores the value of data science in solving complex industrial challenges.