How is the protein separation efficiency of an air classification system measured and validated?

Dry fractionation via air classification has emerged as a sustainable, chemical‑free route to produce pulse and bean protein concentrates, separating protein‑rich fine particles from starch‑dominant coarse particles purely through aerodynamic forces. To guarantee consistent protein enrichment performance for industrial plant‑based protein production, air classification systems rely on standardized measurement workflows and multi‑layer validation procedures to quantify protein separation efficiency (PSE). This article outlines core metrics, laboratory testing routines, process‑oriented validation and full‑scale industrial verification of air‑classifier protein separation performance.

Core Quantitative Metrics for Measuring Separation Efficiency

Protein separation efficiency cannot be judged by particle‑size data alone. Operators combine mass‑balance accounting, proximate chemical analysis and particle‑behaviour curves to evaluate real separation performance.

1. Protein Separation Efficiency (PSE) and Protein Recovery

PSE is the primary indicator for dry pulse‑protein air‑classification performance, calculated through mass balance across input feedstock, fine protein‑enriched fraction and coarse starch‑rich fraction.

PSE = (Protein content of fine fraction × mass yield of fine fraction) ÷ Protein content of original feed flour (dry‑base)

Protein recovery represents the percentage of total input protein successfully captured in the fine product stream. Meanwhile, complementary indicators include:

• Protein enrichment factor: Ratio of fine‑fraction protein content versus raw‑flour protein content, showing how much protein concentration is elevated by classification.
• Mass yield: Weight percentage of fine and coarse fractions recovered from total feed input; total material recovery should normally exceed 95 % for well‑tuned industrial systems.

All protein content values are measured by Kjeldahl nitrogen analysis on moisture‑corrected dry‑basis samples of raw feed, fine output and coarse reject streams. Without full mass balance and chemical testing, particle‑size results cannot reflect true protein‑starch separation effect.

2. Particle‑size metrics and Tromp selectivity curve

Laser diffraction particle‑size analysis characterizes particle‑size distribution (PSD) of every material stream, reporting D10, D50 and D90 values. For pulse‑bean dry fractionation, ultra‑fine grinding typically targets D90:10‑65 μm to liberate protein bodies from starch granules before classification.

The Tromp curve (selectivity curve) is the gold‑standard graphical tool for air‑classifier mechanical performance. It plots particle‑size against fraction partition ratio, delivering key parameters:

• Cut‑point (d50): The particle size with 50 % probability to enter fine or coarse stream, the core tuning target for aerodynamic control.
• Sharpness index: Calculated as D25/D75, quantifies separation sharpness; higher values indicate less cross‑contamination between protein‑rich fines and starch‑rich coarse fractions.
• Apparent bypass fraction: Quantifies undesired fine protein particles mistakenly carried into coarse discharge, representing protein loss and economic waste.

Tromp curves describe mechanical particle sorting capability, while chemical mass‑balance PSE translates particle‑level behaviour into real protein‑separation outcome for pulse raw materials.

Step‑by‑Step Laboratory and Pilot‑Scale Measurement Workflow

Measurement follows a standardized test sequence for pulse raw materials such as pea, mung bean and lentil, matching the full pre‑treatment sequence of commercial dry‑fractionation lines.

1. Standardized sample pre‑treatment: Raw legume materials go through cleaning and precision dehulling to remove hull fibre, producing de‑hulled feedstock identical to industrial input conditions.
2. Controlled ultra‑fine grinding: Dehulled beans are milled to target PSD (D90:10‑65 μm). Consistent grinding is critical: insufficient milling fails to liberate protein‑bodies from starch granules; over‑milling causes excessive starch fragmentation that degrades separation sharpness.
3. Controlled air‑classification trial: Fixed feed rate, classifier‑wheel frequency, airflow velocity are set for the test run. Both fine (protein‑concentrate) and coarse (starch‑rich) outputs are fully collected and weighed for mass‑balance calculation.
4. Sample sampling and chemical testing: Representative aliquots of input flour, fine fraction and coarse fraction are sampled for moisture and total‑protein determination.
5. Metric computation: Calculate mass yield, protein recovery, PSE and enrichment factor. Particle‑size data and Tromp curve are generated from laser‑diffraction measurement on each stream.
6. Parameter sweep tests: Multiple runs are performed by adjusting wheel speed, feed‑rate and air velocity to map performance change against operating parameters, identifying the optimal cut‑point for a given pulse variety and origin.

Multi‑dimensional Validation Approaches

Measured efficiency data needs multi‑level validation to confirm real‑world reliability, covering analytical cross‑checking, microscopic characterization, repeatability testing and raw‑material adaptability assessment.

1. Mass‑balance closure validation

A basic validation check is total mass and protein mass closure. Sum of protein mass contained in fine and coarse fractions should closely match total protein mass fed into the classifier. Significant deviation indicates sampling error, material leakage or product loss during collection, and invalidates the efficiency result.

2. Micro‑morphology validation via electron microscopy

Scanning electron microscopy (SEM) offers visual validation: fine fractions should show abundant small protein‑body fragments, while coarse fractions are dominated by intact large starch granules. If numerous large starch particles appear in fine fraction or tiny protein fragments accumulate in coarse discharge, it visually confirms poor separation sharpness, consistent with low sharpness‑index from Tromp curve testing.

3. Repeatability and reproducibility validation

• Intra‑batch repeatability: Replicate classification runs using identical raw material and identical machine settings. Acceptable industrial tolerance requires PSE variation below ±3 % across repeated trials.
• Inter‑batch reproducibility: Test multiple batches of the same pulse variety from different harvest lots, to verify whether separation performance remains stable against minor natural raw‑material fluctuation.

4. Raw‑material adaptability validation

Air‑classification performance is highly sensitive to pulse variety, growing origin, protein content and seed microstructure. Validation programmes test different beans (pea, mung‑bean, lentil, faba‑bean) to confirm how the system adjusts cut‑point, rotor frequency and air velocity to maintain acceptable PSE across diverse feedstock, matching multi‑parameter intelligent‑optimization functions on commercial dry‑fractionation platforms.

Full‑Scale Industrial‑Site Validation

Laboratory‑pilot results do not always directly translate to large‑scale production. Industrial validation on full production lines completes the final verification for air‑classification systems for pulse‑protein fractionation.

1. On‑line sampling across continuous production: Grab samples are periodically taken from feed inlet, fine‑product outlet and coarse discharge under steady‑state continuous running, avoiding batch‑test bias of small‑scale equipment.
2. Real‑time parameter correlation: Record live operational data: classifier‑wheel frequency, feed throughput, air velocity. Correlate these process readings against offline laboratory PSE results to build production‑grade operating windows. Modern dry‑fractionation equipment supports real‑time multi‑parameter tuning to stabilise protein enrichment for raw‑material variations from different geographical origins.
3. Long‑run stability validation: Perform multi‑day continuous production trials. Monitor drift in PSE, protein content of fine concentrate and mass yield. Rising protein loss into coarse stream indicates issues such as classifier‑wheel fouling, powder agglomeration or filter clogging that degrade long‑term separation efficiency.
4. End‑product functional‑property cross‑check: Beyond composition metrics, validate functional integrity of protein concentrates. Well‑executed dry air‑classification preserves native protein functionality (solubility, emulsification, foaming properties). Severe drop in functional attributes may point to excessive frictional heat or over‑grinding even when numerical PSE values appear favourable.

Common Sources of Measurement Deviation

Several factors can distort measured protein‑separation‑efficiency values and should be controlled during testing and validation:

1. Incomplete dehulling: residual seed hull fibre skews both particle‑size and protein‑content results.
2. Improper grinding: insufficient cell disruption prevents liberation of protein and starch bodies.
3. Poor particle dispersion inside classifier chamber: agglomerated powder causes fine‑protein particles to report incorrectly to coarse fraction, raising bypass fraction.
4. Unstable feed‑rate or airflow fluctuation during trials.
5. Biased sampling: non‑representative subsampling of heterogeneous powder streams.

Measuring and validating protein‑separation‑efficiency for air‑classification systems is a multi‑faceted work, combining chemical mass‑balance calculation of PSE and protein recovery, particle‑size and Tromp‑curve mechanical characterization, microscopic visual evidence, repeatability testing and final full‑scale industrial‑line verification.

Laboratory pilot tests define theoretical performance envelopes; real‑world industrial validation confirms that target separation efficiency can be sustained under continuous production with variable commercial pulse feedstocks. Only by integrating chemical composition, particle‑mechanical behaviour, micro‑structural observation and production‑scale trials can equipment developers and plant operators reliably assess true protein‑fractionation performance of dry air‑classification systems for sustainable chemical‑free plant‑protein manufacturing.