Natural Gas Filters: Interpret Dust Breakthrough as Loading and Face Velocity Change

Published on 11.19

Natural Gas Filters: Interpret Dust Breakthrough as Loading and Face Velocity Change

Abstract

A clean-element efficiency does not fully describe a filter as dust accumulates. Chi and colleagues investigated this changing response in a 2007 study of glass-fibre elements intended for natural-gas filtration. Their laboratory used air and talc at ambient conditions, with optical particle counting. This analysis follows the apparatus, results and conclusions, preserving the eleven source figures and comparative table. It distinguishes outlet concentration, particle-size distribution, fractional efficiency and cake pressure drop. The results connect loading and face velocity with filtration behaviour, including increased penetration of larger particles under some conditions. Their numerical recommendations remain specific to the tested material and configuration; they do not establish a universal gas-filter velocity limit or field replacement interval.

Background and research objective

As a fibrous element loads, its collection mechanism and pressure drop change. A buyer who sees only a clean-filter percentage may therefore lack the information needed to evaluate downstream particle exposure during service. The research question is how loading, face velocity and environmental conditions affect that response.
The source concerns natural-gas filter elements, but the experimental medium was air with talc [1]. This distinction is retained throughout. The study cannot be described as a high-pressure natural-gas station trial, and its loading history cannot directly supply a field service life.

Experimental apparatus and method

Source Figure 1 shows the suction-based test arrangement. The authors used a PALAS Welas 2000 optical particle counter and describe dilution by a factor of ten, with a dust-feeding range of 10–6,000 g/h. Figure 2 presents the talc's volume-based particle-size distribution. The reported refractive-index value is 1.59.
These details matter because particle material and the measurement method influence the interpretation of concentration and size. An optical distribution should not be silently relabelled as a gravimetric result. The source's total mass efficiency and its size-resolved observations answer related but different questions.
One loading analysis used a 16 mm filter layer, a 1 m length, face velocity of 0.02 m/s, inlet concentration of 2 g/m³ and relative humidity of 45%. The velocity comparisons used a 10 mm layer and 464 mm length at 45% relative humidity, with a terminal pressure drop of 20 kPa. Those configurations must remain distinct when comparing figures.
Source Figure 1. Original air/talc filter apparatus, including sampling points, dilution, optical sensors and vacuum flow. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 88.
Original air/talc filter apparatus, including sampling points, dilution, optical sensors and vacuum flow.
Source Figure 2. Original talc volume-frequency and cumulative-volume particle-size distributions. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 88.
Original talc volume-frequency and cumulative-volume particle-size distributions.

Results: loading and collection mechanism

Figure 3 relates outlet concentration to deposited mass per unit filter area. Figure 4 compares the cumulative fraction above particle diameter at different outlet concentrations. Figure 5 tracks cake pressure drop as loading develops. Together, they describe the transition from depth collection towards cake filtration.
The authors explain that capture and resistance evolve together. Increasing pressure drop is therefore not, by itself, proof that the filter has lost collection efficiency. Equally, improved collection during cake formation does not establish that the pressure loss is acceptable for the gas system. Both quantities require their operating context.
The original curves are retained as curves. Without their underlying numerical series, individual points should not be reconstructed as supposedly exact measurements or replaced with fitted data. This is especially important when discussing a change in penetration over the loading history.
Source Figure 3. Original outlet dust concentration versus mass loading, with measured points and source curve unchanged. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 89.
Original outlet dust concentration versus mass loading, with measured points and source curve unchanged.
Source Figure 4. Original oversize particle fractions at three reported outlet concentrations. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 89.
Original oversize particle fractions at three reported outlet concentrations.
Source Figure 5. Original cake pressure drop versus deposited mass per unit area. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 89.
Original cake pressure drop versus deposited mass per unit area.

Results: face velocity, humidity and thickness

Figures 6 and 7 compare particle-size response and large-particle capture with face velocity. Their stated comparison pressure drop is 3 kPa, distinct from the loading test's terminal 20 kPa. Figure 8 presents the associated cake-pressure-drop behaviour. Source Table 1 reports:
U (m/s)
Inlet Cu (mg/m³)
Outlet Cd (mg/m³)
Mass efficiency (%)
0.02
2527.9
0.422
99.983
0.04
4921.8
1.060
99.978
0.06
1033.7
0.364
99.965
0.08
1297.0
1.630
99.874
0.12
1117.8
6.000
99.463
The inlet concentration differs between rows. This table must not be presented as a comparison in which velocity alone changed while every other input remained identical. It also shows why a high overall percentage does not replace examination of outlet concentration and particle size.
Figure 9 compares fractional efficiency at different relative humidities. Figure 10 preserves the clean-element fractional-efficiency comparison, and Figure 11 compares loading behaviour for 7.5 and 12.5 mm thicknesses. Clean efficiency, size-resolved efficiency and overall mass efficiency remain separately identified.
Source Figure 6. Original oversize-fraction curves at five face velocities; the comparison uses the source pressure-drop condition. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 90.
Original oversize-fraction curves at five face velocities; the comparison uses the source pressure-drop condition.
Source Figure 7. Original collection efficiency at three particle diameters versus face velocity. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 90.
Original collection efficiency at three particle diameters versus face velocity.
Source Figure 8. Original cake pressure-drop loading curves for three face velocities. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 90.
Original cake pressure-drop loading curves for three face velocities.
Source Figure 9. Original fractional-efficiency comparison at 45% and 54% relative humidity. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 90.
Original fractional-efficiency comparison at 45% and 54% relative humidity.
Source Figure 10. Original clean-element fractional-efficiency curves at three face velocities. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 91.
Original clean-element fractional-efficiency curves at three face velocities.
Source Figure 11. Original outlet-concentration loading comparison for 7.5 and 12.5 mm filter thicknesses. Chi Huachang, Ji Zhongli, Wei Rongfang (2007), printed p. 91.
Original outlet-concentration loading comparison for 7.5 and 12.5 mm filter thicknesses.

Discussion and the problem resolved

The study turns a broad efficiency claim into a set of reviewable conditions: medium, particle distribution, loading, face velocity, humidity and pressure drop. The authors' recommendation to keep velocity below 0.06 m/s belongs to the tested talc and glass-fibre configuration. It is not a general design maximum for all natural-gas filters.
Field dust composition, gas pressure, droplets and changing demand may differ from the laboratory conditions. The article provides no service replacement time for those combinations. A maintenance interval requires evidence from the actual installation and its approved operating limits.
For procurement, provide flow and pressure ranges, expected particulate or droplet loading, the downstream equipment to protect, allowable pressure drop and maintenance access. Request an efficiency description with a named method and size metric. Comparing two unqualified “99%” statements can conceal different test conditions and different downstream exposure.

Conclusions

The experiments show that loading, face velocity and humidity affect filtration response together. Interpreting breakthrough requires outlet concentration, size-resolved capture and pressure drop to be read within the same test context. One total-efficiency value cannot establish filter lifetime.

Limitations and outlook

The experiments used air, talc and specified elements, without validating every field gas or contaminant. Further evaluation should match the intended conditions and document loading history, repeatability and uncertainty. The evidence can guide an application discussion, while qualification of a proposed product remains separate.

Notes

PDF pages 1–5 contain the target article, with its reference continuation below “continued from page 91” on PDF page 6. The unrelated upper article on that page is excluded. All eleven figures and Table 1 are in scope; the source has no numbered equations. The table values above are transcribed source data, not supplier measurements.

References and Image Sources

[1] Chi Huachang, Ji Zhongli, Wei Rongfang. Experimental study of dynamic filtration performance of a natural-gas fibrous filter. Journal of China University of Petroleum (Edition of Natural Science), 2007,31(6):87–91,97. Complete target text consulted. Original Figures 1–11 and Table 1 belong to the source authors. Public issue listing with the article title and authors: https://www.paper.edu.cn/qk/academic/journal/NUjGkFwQ?cycle=6&year=2007 . The full text was read from the archived original PDF.
Original body visuals: Figure 1 (printed p. 88), Figure 2 (printed p. 88), Figure 3 (printed p. 89), Figure 4 (printed p. 89), Figure 5 (printed p. 89), Figure 6 (printed p. 90), Figure 7 (printed p. 90), Figure 8 (printed p. 90), Figure 9 (printed p. 90), Figure 10 (printed p. 91), Figure 11 (printed p. 91). Visuals credited to [1]. English annotations and cropping were applied to the source excerpts. Reproduction authorised for this publication.
Cover photograph: Tima Miroshnichenko, Pexels. Illustrative stock photograph. https://www.pexels.com/photo/a-man-in-white-coat-holding-an-erlenmeyer-flask-5427858/ . License: https://www.pexels.com/license/ .