Xi'an Yanshuo Instrument Equipment Co., Ltd
Xi'an Yanshuo Instrument Equipment Co., Ltd
Verified Business License Business License
Main Products: Fabric Strength Tester, Friction Colour Fastness Tester
Home > Blog > Pore Size Distribution Testing of Nonwoven Fabrics: From Traditional Sieving to AI Image Analysis

Contact Us

Mrs. Liu
Chat Now

Your inquiry content must be between 10 to 5000 characters

Please enter Your valid email address

Please enter a correct verification code.

Pore Size Distribution Testing of Nonwoven Fabrics: From Traditional Sieving to AI Image Analysis

I. Why is Pore Size Distribution So Important?

The pore size and distribution of nonwoven fabrics directly affect:

Air permeability: Medical protective clothing requires a balance between barrier properties and comfort (optimal pore size: 20-50μm)

Filtration efficiency: The interception rate of melt-blown fabric for PM2.5 is positively correlated with 1-5μm pores

Liquid penetration: When the surface pore size of hygiene products exceeds 40μm, backflow is prone to occur

Industry pain point: Traditional testing methods take up to 4 hours per sample, while AI can reduce the cycle to 15 minutes

II. Traditional Testing Methods vs. Modern Technologies

 

Method

Principle

Accuracy

Limitations

Dry sieving (GB/T 14799)

Vibratory sieving of glass beads

±10μm

Cannot measure micropores <20μm

Capillary flow (ASTM F316)

Bubble point pressure calculation

±2μm

Only reflects the largest through-pore

Mercury intrusion

High-pressure mercury filling

±0.5μm

Destroys samples; cost >$500/test

AI image analysis

SEM image segmentation via deep learning

±0.1μm

Requires 2,000+ labeled training samples

 

Case study: A filter material manufacturer misjudged pore size using traditional sieving (labeled 25μm, actual 38μm), causing mask BFE to drop from 98% to 82%.

III. Three Breakthroughs of AI Image Analysis

1. High-precision pore identification

U-Net algorithm segments SEM images with 96% accuracy (vs. 70% for traditional thresholding)

Can distinguish overlapping pores (red-marked areas in Fig. 1 show algorithm corrections)

https://example.com/sem_ai.png

2. Dynamic pore mapping

Combines X-ray micro-CT (0.7μm resolution) to build 3D pore models

Predicts air permeability with <5% error (vs. GB/T 5453 measured values)

3. Intelligent process feedback

Huawei Cloud EI's real-time monitoring system:

1°C increase in hot-air process temperature → 3.2μm increase in average pore size → automatic fan speed adjustment

Defect detection rate improved from 78% to 99.5%

IV. The Hidden Relationship Between Air Permeability and Pore Size

Theoretical model:

 

 

(Q: Airflow; r: Pore radius; n: Pores per unit area)

 

Empirical data:

Pore size (μm)

Pore density (pores/cm²)

Air permeability (L/m²·s)

15±2

1,200

250

30±3

800

580

50±5

400

920

 

Anomaly: When pore size >60μm, fiber entanglement creates "pseudo-pores," reducing actual permeability by 23%.


Share

Recently Posted

Contact Us

Send Inquiry to Us
* Message
0/5000

Want the best price? Post an RFQ now!

Recommended Products