Low Moisture Extrusion Dataset for Six Plant Protein Isolates and Binary Blends in Textured Vegetable Protein (TVP) Production

Full description

This collection contains experimental data generated from a comparative study of six commercially relevant plant protein ingredients (soy, pea, chickpea, faba bean, lentil, and mung bean) processed via low moisture extrusion for dry textured vegetable protein (TVP) production.

The dataset includes ingredient characterisation (RVA pasting behaviour, colour, water absorption capacity), extrusion process conditions (feed rate, screw speed, water flowrate, in-barrel moisture content, motor torque and specific mechanical energy), and extrudate properties (bulk and apparent density, expansion ratio, porosity, water holding capacity, colour, and texture (per-piece texture profile analysis recomputed from the raw force–deformation curves)).

Binary 50:50 protein blends were also evaluated across multiple moisture levels to examine formulation–process interactions.

The data were produced to support systematic comparison of protein-specific extrusion behaviour and to inform formulation-specific process optimisation strategies in plant-based meat analogue development. The dataset enables reuse for comparative analysis, modelling, meta-analysis, and process optimisation studies in food extrusion research.
Lineage: Data were generated through pilot-scale twin-screw extrusion trials conducted at the CSIRO Food Innovation Centre (Werribee, Victoria, Australia) using a Clextral Evolum 32 co-rotating twin-screw extruder.

Six commercial protein ingredients were first evaluated individually to determine feasible and stable extrusion conditions capable of producing acceptable dry TVP structures. Processing parameters (feed rate, screw speed, water flowrate, barrel temperature profile) were adjusted per protein to define protein-specific processing windows.

Subsequently, 50:50 binary blends were extruded across a controlled range of in-barrel moisture contents under a standardised temperature profile.

Ingredient characterisation included:

Rapid Visco Analysis (RVA) pasting properties

Water absorption capacity (centrifugation method)

Colour (CIELAB measurements)

Extrudate characterisation included:

Bulk, apparent, and true density

Porosity (derived from density measurements)

Expansion ratio

Water holding capacity

Texture profile analysis (double compression test)

Replicates were performed for ingredient and extrudate measurements as described in the associated publication.

This collection corresponds to the dataset supporting the manuscript:
“Blending Shifts the Processing Envelope for Low Moisture Extrusion of Six Legume Proteins”

Version 2 (October 2026). Two files were added: 6Proteins_ProcessConditions.csv, giving specific mechanical energy, motor torque, feed rate, water flowrate and in-barrel moisture content for all 56 extrusion runs across both trials, including the 25 runs that were extruded but yielded no characterisable product; and 6Proteins_Texture_recomputed.csv, giving per-piece texture profile analysis for all 299 archived force–deformation curves, recomputed from the raw curves rather than taken from the instrument software. Version 1 contained neither, and the processing envelope and texture results could not be reproduced from it.

One error was corrected: the mung bean replicate 2 pasting curve in 6Proteins_RVA.csv was a duplicate of the faba bean replicate 1 curve, and has been replaced with the correct curve from the instrument records. Two metric labels in 6Proteins_Trial2_alldata.csv were also corrected: Ture to True and Resilence to Resilience.