A coalescing filter and a downstream mist eliminator can both collect liquid, but their shares of collected liquid do not have the same denominator as overall separation efficiency. Li, Ji and Chen used a negative-pressure air bench with a low-volatility liquid to examine that distinction. The study weighed the injected liquid and the liquid collected from the housing wall, cartridge and mesh. This article follows the experiment and preserves the three mass ratios. It explains why a small mesh share does not quantify liquid at the outlet and why stage performance must be compared at matched droplet size, loading and flow conditions.

A natural-gas filter separator may need to deal with solid particles and entrained liquid. A coalescing cartridge and a mesh element do different jobs in a staged separator. To compare offers, a buyer has to know whether a reported percentage refers to all liquid fed into the rig, liquid captured by one element, or liquid still carried out of the assembly. A number without a denominator can lead to a wrong interpretation even if the original experiment was clear.

Li, Ji and Chen developed a bench modeled on a horizontal natural-gas filter [1]. This review asks what the weighing method can distinguish, what the velocity, droplet and loading comparisons actually show, and which measurements a buyer still needs to evaluate an offered assembly.

Figure 1. Air/DOS apparatus with feed instrumentation, differential pressure measurements, collection bottles and outlet suction. Li, Ji and Chen [1], original Figure 1 (print p. 123). Functional English redraw. No dimensions inferred from the drawing.

The experiment used an atomization system, a separation assembly and an air supply under negative pressure. Compressed air and a liquid feed entered an atomizing nozzle. The paper describes measuring the droplet distribution with a particle analyzer. The first transparent cylinder contained a glass-fiber cartridge with an effective filtration length of 460 mm and an outer diameter of 86 mm. The second cylinder contained a metal mesh mist eliminator. U-tube manometers measured the pressure drop across the cartridge and mesh separately.

The liquid was dioctyl sebacate (DOS), with a reported density of 913 kg/m³ and viscosity of 25 mPa·s at 20 °C. The authors treated evaporation at room temperature as negligible. They conditioned a cartridge that already contained liquid until no additional liquid was drawn out and its pressure drop was stable. The bench used air and DOS, so the results characterize those test conditions. A different gas density, liquid or operating pressure may change the performance of the same separation geometry.

The authors weighed the introduced liquid W. Separate bottles collected the front housing-wall liquid W₁, the liquid drained from the coalescing cartridge W₂, and the mesh-collected liquid W₃. Their definitions are:

η compares captured mass with injected mass. p₁ and p₂ distribute captured mass among locations. For example, in the velocity series the reported mesh share ranged from 0.3% to 0.7% of the collected mass. That percentage does not tell the reader the outlet liquid loading. Even the difference between injected and collected mass cannot be assigned exclusively to outlet carryover without accounting for liquid stored in the apparatus and other measurement effects.

Figure 2. Cartridge pressure-drop histories for four face velocities. Li, Ji and Chen [1], original Figure 2 (print p. 124). Approximate trace digitized from the printed plot. The original measurement file was unavailable.

Figure 3. Filter efficiency and share of collected liquid recovered from the cartridge versus face velocity. Li, Ji and Chen [1], original Figure 3 (print p. 124). Approximate trace digitized from the printed plot. The original measurement file was unavailable.

Figure 4. Share of total collected liquid recovered from the downstream mesh versus face velocity. Li, Ji and Chen [1], original Figure 4 (print p. 124). Approximate trace digitized from the printed plot. The original measurement file was unavailable.

Figure 5. Cartridge pressure-drop histories for four mean droplet diameters. Li, Ji and Chen [1], original Figure 5 (print p. 124). Approximate trace digitized from the printed plot. The original measurement file was unavailable.

Figure 6. Filter efficiency and cartridge liquid share versus droplet diameter. Li, Ji and Chen [1], original Figure 6 (print p. 125). Approximate trace digitized from the printed plot. The original measurement file was unavailable.

Figure 7. Cartridge pressure-drop histories for four DOS aerosol loadings. Li, Ji and Chen [1], original Figure 7 (print p. 125). Approximate trace digitized from the printed plot. The original measurement file was unavailable.

Figure 8. Filter efficiency and cartridge liquid share versus DOS aerosol loading. Li, Ji and Chen [1], original Figure 8 (print p. 125). Approximate trace digitized from the printed plot. The original measurement file was unavailable.

In the face-velocity series the authors used an inlet liquid concentration of 18.5 g/m³ and a mean droplet diameter of 20 μm. They reported that pressure drop rose slowly at first, then rose more quickly before leveling. They interpreted this as liquid progressively accumulating in the cartridge until drainage and entrainment approached a balance. Both the overall collection efficiency and the cartridge share increased with velocity within that series. The mesh still collected some liquid. Its small share is consistent with its downstream position in that bench, and does not establish that a mesh would be unnecessary for a different droplet population.

The droplet-size series kept the face velocity at 0.10 m/s and loading at 11 g/m³. The paper reports that increasing mean droplet size raised overall collection efficiency while the cartridge's share of collected liquid tended to fall. The loading series kept the velocity at 0.10 m/s and mean diameter at 20 μm. Increasing liquid loading again raised overall efficiency and reduced the cartridge share in the tested range. A change in total capture can coexist with a change in where that liquid is collected. These comparisons should be read as separate series rather than pooled into one performance curve.

The report's time series also matters to a comparison. A pressure drop read near the start of liquid loading represents a different cartridge state from a reading after drainage has stabilized. The published curves describe those states for the test rig. A laboratory comparison can state the preconditioning procedure, sampling interval and criterion for a stable result, then report initial and stabilized readings separately. Otherwise, two pressure-drop figures taken at the same nominal flow can describe different liquid inventories in the element.

When comparing a natural-gas filter separator, request the inlet liquid loading and droplet-size distribution, gas conditions, pressure drop by stage, collected mass by location and outlet carryover measurement. The bench illustrates how to define these measurements and interpret their ratios. To extend the results to pressurized natural-gas service, use representative gas properties and liquid behavior and measure the outlet independently. This would make it possible to separate overall capture, the contribution of each collection location and any liquid leaving the separator.

All ratios above are dimensionless. Multiply by 100 to express them as percentages. The reported 0.3%–0.7% is p₂ in the velocity series. It describes the cited rig rather than a commercial separator rating.

[1] Li, B., Ji, Z., and Chen, H. (2007). “Experimental study of gas–liquid separation performance of natural gas filters” [Experimental study of gas–liquid separation in natural-gas filters]. Natural Gas Industry, 27(10), pp. 123–125. Sections I–IV and Figures 1–8. Figure 1 is a functional redraw. Figures 2–8 show approximate readings from the published plots. https://www.sciengine.com/parse/pdf/1000-0976/89E36EC2A19044F79A2867261AF6E9F9.pdf

Cover photo: U.S. Department of Energy, Oak Ridge archive. Oak Ridge health physicist Max Scott positioning a sample at a gamma counter. U.S. Government work, public domain per the source record. Editorial photo crop and title layout. https://commons.wikimedia.org/wiki/File:HD.5B.026_(11874519294).jpg