Chamber-to-Field Development and Validation of Integrated Filtration–Adsorption Air Cleaners for Particle and VOC Control in Shipboard Cabins

Abstract

Shipboard cabins can contain both oil mist-dominated particulate matter and volatile organic compounds (VOCs), creating a need for compact air-cleaning systems capable of simultaneous particle and gas-phase removal. This study developed an integrated filtration–adsorption air cleaner through a performance-oriented route combining a low-resistance gradient filter with an activated-carbon packed bed. The gradient filter maintained PM_2.5 single-pass removal efficiencies above 99.99% across airflow rates of 150–650 m³/h and under inlet concentrations up to approximately 11 mg/m³, while maintaining relatively low airflow resistance. Three activated carbons were also systematically compared based on particle morphology, packed-bed characteristics, pore structure, and dynamic adsorption performance. JZT-07 showed the most favorable overall characteristics and achieved breakthrough capacities of 132, 167, and 159 mg/g for benzene, toluene, and n-heptane, respectively. The selected filtration and adsorption modules were integrated into accommodation- and machinery-cabin air cleaners and evaluated under standard-chamber and shipboard conditions. During treated voyages, average PM₁₀ and TVOC concentrations decreased by 75.3% and 66.4% in the accommodation cabin and by 65.8% and 69.6% in the machinery cabin, respectively. These results demonstrate a practical development route from component optimization to device integration and field validation for simultaneous particulate and VOC control in confined ship environments.

1. Introduction

Ship cabins represent a demanding indoor-air environment because particulate matter and volatile organic compounds (VOCs) can be generated simultaneously within confined spaces [1,2,3]. Machinery operation [2], lubricating oils [4], and human occupancy can continuously introduce both particle-phase and gas-phase pollutants, while ventilation capacity, fan power, operating noise, and maintenance access remain constrained. Under these conditions, an air-cleaning system must provide high pollutant-removal performance without imposing excessive airflow resistance or device burden.

Oil-related particulate pollution is particularly challenging because mechanical atomization, leakage, splashing, evaporation, and condensation can generate liquid or semi-liquid aerosols over a broad particle-size range [4]. Fine particles can remain suspended for long periods and penetrate into occupied areas, making high-efficiency filtration necessary [5,6]. However, improving filtration efficiency by using finer fibers, denser media, or thicker filter layers generally increases pressure drop and therefore places a greater airflow demand on the fan [7,8,9]. For compact shipboard devices, this efficiency–resistance trade-off is critical [10,11]. A gradient filtration structure provides a practical route to distribute particle loading and filtration duty across different layers, allowing a relatively open upstream layer to intercept larger particles while a finer downstream layer provides the required PM_2.5 removals.

Although particulate matter can be effectively removed by the upstream filtration stage, gas-phase VOCs remain in the airflow [12,13] and require a separate purification mechanism [14,15,16]. Activated carbon is widely used for VOC removal [17,18] because of its developed pore structure and broad affinity toward organic compounds. Nevertheless, the suitability of an activated carbon for a packed-bed air cleaner cannot be judged by BET surface area alone [19]. Particle morphology affects packing density and bed void fraction, while micropore surface area and micropore volume determine the availability of adsorption sites. These characteristics jointly influence adsorption capacity [20], gas transport, and packed-bed resistance [21,22,23]. In addition, real oil-derived atmospheres contain multiple VOCs that compete for adsorption sites. Competitive adsorption can alter breakthrough order [24,25] and may even displace previously adsorbed compounds, so adsorbent selection should be based on dynamic breakthrough behavior under both single-component and multi-component conditions [26].

The particle and gas phases therefore require different but coordinated purification mechanisms. A high-efficiency particle filter must operate without consuming excessive pressure allowance, while the downstream adsorbent must provide sufficient VOC capacity without introducing excessive packed-bed resistance. The two components must operate in series within a common airflow path, and the resulting device must still provide adequate airflow, clean-air delivery, acceptable power demand, and manageable noise. Moreover, satisfactory component-level or standard-chamber performance [27,28,29] does not necessarily guarantee effective pollutant control during actual voyages [3,30,31], where pollutant sources, ventilation conditions, and operating states vary continuously [32]. Therefore, the practical development of shipboard air cleaners requires not only optimization of individual purification components, but also validation across material, device, chamber, and field scales [30].

Accordingly, this study adopted a performance-oriented, chamber-to-field development route for integrated filtration–adsorption air cleaners. The particle-removal module was first designed around explicit requirements for airflow, PM_2.5 removal efficiency, and pressure drop, and its performance was evaluated over a broad range of airflow rates and inlet particle concentrations. Candidate activated carbons were then compared in terms of particle morphology, packed-bed characteristics, pore structure, and dynamic adsorption behavior toward benzene, toluene, and n-heptane [21]. Single-component and ternary breakthrough tests were used to identify an adsorbent capable of maintaining strong performance under competitive VOC conditions [33]. The selected gradient filter and activated carbon were subsequently integrated into two air-cleaner configurations designed for accommodation and machinery cabins, followed by standard-chamber evaluation and multi-voyage shipboard validation. This work establishes a systematic route from component-level performance requirements to adsorbent screening, device integration, and field verification [27], providing an engineering framework for simultaneously balancing particle removal, VOC adsorption, airflow resistance, and practical applicability in confined ship environments [34].

2. Materials and Methods

2.1. Materials and Air-Cleaner Configurations

This study developed two modular ship-cabin air cleaners for the integrated control of oil-mist particles and VOCs. The design concept and prototypes are shown in Figure 1. Both air cleaners were based on the same serial purification route: contaminated cabin air first passed through an upstream gradient fibrous filter module for particle removal and then entered a downstream activated-carbon packed bed for gas-phase VOC adsorption. The two units differed mainly in scale and intended application. The Type A air cleaner shown in Figure 1b was designed for accommodation cabins, where compactness and low-noise operation are important, whereas the Type B air cleaner (Figure 1c) was designed for machinery cabins, where higher airflow and stronger pollutant-removal capacity are required.

Figure 1. Design concept and prototypes of a two-stage ship-cabin air cleaner for integrated removal of oil-mist particles and VOCs. (a) Schematic of the purification pathway, consisting of a gradient filter module for oil-mist particle removal and a JZT-07 activated-carbon packed bed for VOC adsorption. Within the gradient filter schematic, the light-blue, relatively sparse fibrous region represents the coarse filter layer for capturing larger oil-mist particles, while the darker-blue, denser fibrous region represents the high-efficiency filter layer for further removal of fine particles. (b) Type A air cleaner for accommodation cabins. (c) Type B air cleaner for machinery cabins.

2.1.1. Overall Configuration of the Air Cleaners

The air cleaners were configured as through-flow devices in which all treatment units were arranged along a single airflow path. As illustrated in Figure 1a, polluted ship-cabin air containing both oil-mist particles and VOCs enters the device from the inlet side. The first treatment stage is the gradient filter module, which removes oil mist-dominated particulate matter before the air reaches the gas-phase adsorption section. The particle-reduced but VOC-containing air then passes through the activated-carbon packed bed, where VOCs are adsorbed by the porous activated-carbon adsorbent. A fan installed downstream of the treatment modules provides the driving force for continuous airflow and discharges the treated air back to the cabin.

This arrangement was selected to match the physical form of each pollutant with the corresponding removal mechanism. Oil-mist particles are removed primarily by fibrous filtration, including interception, inertial impaction, and deposition within the filter layers [35]. VOCs, by contrast, are removed by adsorption within the activated-carbon packed bed. Placing the gradient filter upstream also limits the entry of oil droplets and suspended particles from directly entering the carbon bed, thereby reducing particulate loading, pore blockage, and unnecessary pressure drop in the VOC adsorption module. In this way, the activated-carbon adsorbent is used mainly as a gas-phase adsorbent.

The two functional modules were integrated within a common enclosure but remained conceptually and structurally separable, allowing the particle filter and carbon bed to be inspected, maintained, and replaced independently. The Type A prototype for accommodation cabins had external dimensions of 330 × 300 × 215 mm and a mass of 10 kg, whereas the Type B prototype for machinery cabins was larger, with dimensions of 680 × 340 × 415 mm and a mass of 30.0 kg. These two prototypes followed the same two-stage treatment principle but were scaled for different cabin environments and airflow demands. Their device-level performance was subsequently evaluated through standard-chamber and engineering tests, including ventilation airflow rate, PM_2.5 and xylene clean-air delivery rate (CADR), and operating noise. The corresponding test methods are described in Section 2.2.

2.1.2. Gradient Filtration Module for Particle Removal

The gradient filtration module was designed as the upstream particle-removal stage of the air cleaner. Its primary function was to remove oil-mist particles and other suspended particulate matter from ship-cabin air while maintaining a low initial pressure drop. The module adopted a layered gradient structure so that particle loading could be distributed across different filtration layers. As shown in Figure 1a, the gradient filter stack consisted of a coarse filter layer, a base fabric, a medium-efficiency filter layer, and a high-efficiency filter layer arranged from the inlet side to the outlet side. The upstream coarse layer was used to intercept larger oil-mist droplets and reduce the particle load entering the downstream layers. The base fabric provided mechanical support for the filter stack. The medium-efficiency and high-efficiency layers then supplied the additional capture of fine particles, especially the PM_2.5 fraction, which is more difficult to remove and more likely to penetrate into occupied cabin spaces. This graded configuration was intended to balance particle-removal efficiency, pressure drop, and long-term loading behavior.

In the present air-cleaner design, the gradient filter module also served as a protective unit for the downstream activated-carbon bed. By removing particulate matter before gas-phase adsorption, the filter reduced the risk of oil deposition and dust accumulation on the carbon surface. Therefore, the gradient filtration module was not only responsible for particulate purification but also for maintaining the effective operation of the VOC adsorption module during shipboard use.

2.1.3. Activated-Carbon Module for VOC Adsorption

The activated-carbon adsorption module was installed downstream of the gradient filter and served as the second treatment stage for gas-phase VOC removal. After oil-mist particles were largely removed by the upstream filter module, the remaining VOC-containing air entered the packed carbon bed. This sequencing allowed the activated carbon to operate under a particle-reduced airflow condition.

Three activated-carbon adsorbents with different particle morphologies were considered as candidate materials for the VOC-removal packed bed: cylindrical TX-1, crushed JZT-07, and spherical QZ09-4. Their photographs are provided in Figure S1. These three adsorbents were identified by their sample codes and particle morphologies. TX-1 was a cylindrical activated carbon with a particle size mainly around 0.9 mm, JZT-07 was a crushed activated carbon with relatively coarse and irregular particles, and QZ09-4 was a spherical activated carbon with a more regular particle shape. Before adsorption testing, the candidate adsorbents were characterized in terms of particle-size distribution, packing density, surface morphology, and pore-structure parameters. Based on the screening procedure described in Section 2.2, JZT-07 was selected as the final adsorbent for the VOC-removal module and was packed as a replaceable fixed bed in the integrated air cleaners. These three materials were commercially available engineering adsorbents manufactured by Shanxi Xinhua Chemical Co., Ltd. (Taiyuan, China) and were not synthesized, carbonized, activated, chemically modified, or otherwise prepared by the authors.

For engineering integration, the full-scale rectangular JZT-07 carbon modules were dimensioned according to the available module space within each prototype and the rated airflow requirement of the corresponding cabin application. The effective carbon-bed dimensions were 276 × 174 × 28 mm for the Type A accommodation-cabin unit and 371 × 317 × 28 mm for the Type B machinery-cabin unit, corresponding to effective packed-bed volumes of 1.345 and 3.293 L, respectively. Based on an engineering packing density of approximately 0.40 kg/L, the corresponding JZT-07 loadings were approximately 0.54 and 1.32 kg. Using the rated device airflow rates of 78 and 413 m³/h, the nominal empty-bed contact times were 0.062 and 0.029 s for Type A and Type B, respectively.

The combination of the upstream gradient filter and downstream JZT-07 packed bed therefore formed a two-stage, phase-targeted purification system. The first stage controlled oil-mist particles through fibrous filtration, and the second stage controlled gas-phase VOCs through activated-carbon adsorption. This configuration is the basis for the Type A and Type B air-cleaner prototypes evaluated in the subsequent laboratory, standard-chamber, and shipboard field tests.

2.2. Component and Standard-Chamber Test Methods

The laboratory evaluation was conducted at both component and device levels to support the development of the two-stage ship-cabin air cleaner. The gradient filter module was first tested as the particle-removal unit to evaluate its airflow resistance and single-pass PM_2.5 removal performance. The activated-carbon packed bed was then evaluated through fixed-bed dynamic adsorption tests to screen the VOC adsorbent under single-component and multi-component conditions. Finally, the integrated Type A and Type B air cleaners were evaluated in a standard test chamber to determine their device-level performance, including ventilation airflow rate, PM_2.5 CADR, xylene CADR, operating noise, and power consumption.

2.2.1. Gradient Filter Performance Test

The particle-removal performance of the gradient filter module was evaluated according to the Chinese standard Air Cleaner (GB/T 18801-2022) [36]. The filtration module was installed in a test duct consisting of a PM_2.5 aerosol generation section, a mixing section, a filter test section, and upstream/downstream aerosol sampling ports. The test system was equipped with a positive-pressure cigarette-smoke aerosol generator (PG-A, YIKE, Suzhou, China), a hot-wire anemometer (testo 405i, Testo SE & Co. KGaA, Titisee-Neustadt, Germany), a differential pressure gauge (DP-CALC 5825, TSI Inc., Shoreview, MN, USA), and a PM_2.5 aerosol monitor (DustTrak DRX 8533, TSI Inc., Shoreview, USA).

During the test, PM_2.5 aerosol was generated upstream of the filter module using the aerosol generator and introduced into the test duct together with the carrier airflow. The generated aerosol passed through the mixing section to obtain a stable upstream concentration. The upstream aerosol concentration was adjusted by controlling the aerosol generation rate and the test airflow rate. For each test condition, aerosol samples were taken from the upstream and downstream sampling ports to determine the PM_2.5 concentrations before and after the filter module. The PM_2.5 concentrations were measured using the PM_2.5 aerosol monitor.

Two complementary test series were conducted. In the first series, the filter module was tested at airflow rates ranging from 150 to 650 m³/h, corresponding to face velocities of 0.28–1.20 m/s. The face velocity was calculated from the measured volumetric airflow rate and the effective frontal area of the filter module. The pressure drop was determined from the static-pressure difference between the upstream and downstream sides of the filter. In the second series, the filtration performance was independently evaluated at a fixed airflow rate of 250 m³/h. The upstream PM_2.5 concentration was progressively increased from approximately 0.73 to 11.0 mg/m³, and the corresponding downstream concentration and filter pressure drop were measured. This test was used to evaluate the robustness of the filter under substantially elevated PM_2.5 challenge concentrations. Because the two test series were conducted independently, the pressure-drop values obtained at the same nominal airflow rate represent measurements from separate test campaigns. The single-pass PM_2.5 removal efficiency was calculated as Equation (1).

$${\eta }_{\mathrm{pm}}={\displaystyle \frac{C_{\mathrm{pm},\mathrm{in}}-C_{\mathrm{pm},\mathrm{out}}}{C_{\mathrm{pm},\mathrm{in}}}}\times 100\%$$

(1)

where C_pm,in and C_pm,out are the upstream and downstream PM_2.5 concentrations, respectively, in mg/m³.

2.2.2. Packed-Bed VOC Adsorption Test

Packed-bed dynamic adsorption tests were conducted to screen the activated-carbon adsorbent used in the downstream VOC-removal module. Three activated-carbon adsorbents with different particle morphologies were considered: cylindrical TX-1, crushed JZT-07, and spherical QZ09-4. Before the adsorption tests, their basic physical and textural properties were characterized to support packed-bed design and adsorbent selection. Particle-size distribution was determined according to GB/T 7702.2-1997 [37]. Packing density was measured according to GB/T 7702.4-1997 [38], while true density and the calculated bed void fraction were determined according to GB/T 6155-2008 [39]. Surface morphology was observed using a Nova Nano SEM 450 field-emission scanning electron microscope (FEI Company, Hillsboro, OR, USA). Pore-structure parameters, including BET specific surface area, micropore surface area, total pore volume, micropore volume, and mesopore volume, were determined by nitrogen adsorption according to GB/T 7702.20-2008 [40].

The packed-bed adsorption system consisted of compressed gas cylinders, mass-flow controllers (MC 20SLPM, Alicat Scientific, Tucson, AZ, USA), a humidification unit, a temperature and humidity sensor (COS-03, Shandong Renke Control Technology Co., Ltd., Jinan, Shandong, China), a gas-mixing manifold, a packed-bed adsorption column, and gas chromatography-based VOC analysis instruments. Nitrogen and high-purity air were metered using mass-flow controllers and conditioned through the humidification unit to obtain a relative humidity of 50% RH. The humidified carrier gas was then mixed with the VOC standard gas stream through a three-way valve. The inlet VOC concentration was adjusted by controlling the flow-rate ratio of the gas streams, and a stable 100 ppm inlet concentration of each VOC was verified by gas chromatography before the adsorption test. The tests were conducted at 25 °C and 101.325 kPa. The outlet gas from the adsorption column was sampled and analyzed using a gas chromatography–mass spectrometry system (GCMS-QP2010 Plus, Shimadzu Corporation, Kyoto, Japan). The effective detection limit for VOC analysis was 0.1 ppm.

For each adsorption test, dry activated-carbon samples were packed into a cylindrical adsorption column with an internal diameter of 12 mm. A dry adsorbent mass of 1.0 g was used for each test to enable direct comparison of the mass-normalized dynamic adsorption performance among the candidate materials. The total gas flow rate was maintained at 1200 mL/min. For the single-component tests, the inlet concentration of benzene, toluene, or n-heptane was individually set to 100 ppm. For the ternary tests, benzene, toluene, and n-heptane were simultaneously introduced at 100 ppm for each component, corresponding to a total VOC concentration of 300 ppm. The outlet VOC concentration was measured as a function of adsorption time to construct the breakthrough curve. The packed-bed adsorption tests were organized under two VOC exposure scenarios. First, single-component adsorption tests were performed for benzene, toluene, and n-heptane to compare the basic adsorption capacity and breakthrough behavior of the candidate adsorbents. Second, a ternary benzene–toluene–n-heptane mixture was used as a simplified multi-component VOC model system to evaluate competitive adsorption, displacement, and multi-component breakthrough behavior. This design allowed the candidate adsorbents to be assessed under both idealized single-pollutant conditions and a more complex mixed-VOC condition relevant to oil-derived emissions.

The outlet VOC concentration of each VOC (C_VOC,out, ppm) was directly plotted as a function of adsorption time to obtain the dynamic breakthrough curves. The concentration ratio C_VOC,out/C_VOC,0 was used only to define the characteristic breakthrough and saturation thresholds. Breakthrough was defined as the first time at which C_VOC,out/C_VOC,0 = 0.05, and saturation was defined as the first time at which C_VOC,out/C_VOC,0 = 0.95. For the ternary tests, these criteria were applied independently to each VOC using its corresponding inlet concentration of 100 ppm. The corresponding breakthrough time and saturation time were used to calculate the breakthrough adsorption capacity and saturation adsorption capacity, respectively. The dynamic adsorption capacity was calculated by integrating the inlet–outlet concentration difference as Equation (2).

$$q_{\mathrm{VOC}}={\displaystyle \frac{Q_{\mathrm{air}}{M}_{\mathrm{VOC}}}{w{V}_{\mathrm{M},\mathrm{VOC}}}}{\displaystyle \int (C_{\mathrm{VOC},\mathrm{in}}-C_{\mathrm{VOC},\mathrm{out}})dt}$$

(2)

where q_VOC is the adsorption capacity, mg/g; Q_air is the total gas flow rate, m³/min; w is the dry mass of adsorbent packed in the column, g; M_VOC is the molecular weight of the VOC, g/mol; and V_M,VOC is the molar volume of gas at 25 °C and 101.325 kPa, the value of which was obtained as 24.5 L/mol. For multi-component tests exhibiting competitive displacement and concentration overshoot, the saturation adsorption capacity defined above represents the net adsorption accumulated up to the first 95% outlet-concentration threshold. The final adsorbent for the integrated air cleaners was selected by considering adsorption capacity, breakthrough lifetime, and multi-component competitive adsorption behavior. The laboratory breakthrough tests were therefore used primarily for comparative adsorbent screening and capacity benchmarking rather than as a direct geometric scale-up law; the full-scale module geometry was determined by device-level spatial and airflow constraints.

2.2.3. Standard-Chamber Test of the Integrated Air Cleaners

The integrated Type A and Type B air cleaners were evaluated in a standard test chamber to quantify their device-level performance. Unlike the component-level tests described in Section 2.2.1 and Section 2.2.2, the standard-chamber test was conducted on the complete air-cleaning devices, including the upstream gradient filter module, downstream activated-carbon packed bed, fan, enclosure, and internal airflow pathway. The Type A air cleaner was designed for accommodation cabins, whereas the Type B air cleaner was designed for machinery cabins. Both Type A and Type B air cleaners were evaluated under the same standard-chamber testing framework. The device-level performance tests, including PM_2.5 CADR, xylene CADR, input power, operating noise, and purification energy efficiency, were conducted according to GB/T 18801-2022. The standard test chamber used for CADR evaluation complied with QB/T 5364-2019 [41], which specifies the technical requirements and evaluation methods for air-cleaner test chambers. During the standard-chamber tests, the chamber environment was controlled at 23 ± 2 °C and 50 ± 10% RH. Chamber mixing was maintained according to the standard test-chamber procedure to ensure that the challenge pollutant was uniformly distributed before the concentration decay test.

PM_2.5 and xylene were used as the representative particle-phase and gas-phase challenge pollutants, respectively. The PM_2.5 aerosol generation and mixing procedure followed the method described for the gradient filter module in Section 2.2.1, while xylene was generated and mixed using the gas-preparation procedure described for VOC adsorption testing in Section 2.2.2. For each pollutant, two chamber decay tests were conducted: a natural-decay test without air-cleaner operation and an active-cleaning test with the air cleaner operating under the rated condition. During the PM_2.5 CADR test, the time-dependent chamber PM_2.5 concentration was recorded using a PM_2.5 aerosol monitor (DustTrak DRX 8533, TSI Inc., Shoreview, USA). During the xylene CADR test, the time-dependent xylene concentration was measured using a gas chromatography–mass spectrometry system (GCMS-QP2010 Plus, Shimadzu Corporation, Japan).

Following a data-quality audit, the Type A PM_2.5 CADR dataset was excluded from the quantitative analysis because the archived records did not permit verification that the actual device airflow and chamber operating and mixing conditions had been controlled consistently. The retained chamber datasets were Type A–xylene, Type B–PM_2.5, and Type B–xylene.

The PM_2.5 CADR and xylene CADR were calculated from the corresponding natural-decay and active-cleaning concentration decay curves [42,43], shown as Equation (3).

$$CADR=V_{\mathrm{cham}}(k_{\mathrm{active}}-k_{\mathrm{natural}})$$

(3)

where V_cham is the chamber volume, m³; k_active is the first-order decay coefficient obtained with the air cleaner operating, h⁻¹; and k_natural is the natural-decay coefficient obtained without air-cleaner operation, h⁻¹. The detailed CADR calculation method, including the determination of the decay coefficients from chamber concentration decay data, is provided in Supplementary Section S1. The chamber volume used for all retained CADR calculations was 30.0 m³. The corresponding natural-decay and active-cleaning regression plots are provided in Supplementary Figures S2–S4, and the fitting intervals, fitted k_natural and k_active values, R² values, and recalculated CADRs are summarized in Supplementary Table S2.

The Type A and Type B air cleaners were operated under their rated airflow conditions during the standard-chamber tests. The rated airflow rates were 78 m³/h for the Type A accommodation-cabin air cleaner and 413 m³/h for the Type B machinery-cabin air cleaner. Under these rated operating conditions, the PM_2.5 CADR and xylene CADR were determined from the chamber concentration decay tests, while the input power and operating noise were measured as device-level operating parameters. The input power was measured using a digital power meter (WT310E, Yokogawa Test & Measurement Corporation, Hachioji, Japan) during steady operation at the rated airflow condition. The operating noise was measured using a sound level meter (AWA6228+, Hangzhou Aihua Instruments Co., Ltd., Hangzhou, China) under the same operating condition and reported as the A-weighted sound pressure level, dB(A). The purification energy efficiency can then be calculated as the ratio of CADR to input power, as shown in Equation (4).

$$E_{\mathrm{CADR}}={\displaystyle \frac{CADR}{P}}$$

(4)

where E_CADR is the purification energy efficiency, m³/(h⋅W), and P is the input power of the air cleaner, W. PM_2.5 purification energy efficiency and xylene purification energy efficiency were calculated separately using their corresponding CADR values.

2.3. Shipboard Field Test

Shipboard field tests were conducted to evaluate the practical performance of the two-stage air cleaners under real cabin operating conditions. The Type A accommodation-cabin air cleaner and the Type B machinery-cabin air cleaner were deployed in enclosed cabins of a ship series according to the task requirements and cabin pollution characteristics. The Type A units were installed in accommodation-cabin areas, whereas the Type B units were installed in key pollution areas of machinery cabins. The field deployment targeted the simultaneous control of oil-mist particles and VOCs during ship operation. The air-cleaning devices had been deployed for more than 30 voyages, and selected voyages were monitored to evaluate the field performance of the purification system. The field test was designed as a multi-voyage engineering comparison between untreated voyages and treated voyages. Since shipboard pollutant levels are affected by fuel and oil consumption, machinery operating conditions, ventilation status, and crew activities, the field test was used to evaluate the practical reduction and stabilization of cabin pollutant concentrations under real operating conditions.

During navigation, the devices were operated continuously and the purification cartridges were replaced every 15 days (approximately 360 h per replacement cycle). Periodic operational monitoring indicated that the daily mean VOC and particulate concentrations remained generally stable during each operating period, although dynamic fluctuations occurred with voyage day and changing shipboard conditions. Because the monitored voyages spanned multiple cartridge-replacement cycles, voyage number was used for the untreated-versus-treated engineering comparison and should not be interpreted as the cumulative age of a single activated-carbon bed.

Oil-mist particles were monitored using PM₁₀ mass concentration as the representative particle-phase indicator. During the field monitoring periods, PM₁₀ concentration was continuously measured using a DustTrak DRX 8533 aerosol mass monitor (TSI Inc., Shoreview, USA), with a data-recording interval of 1 min. The PM₁₀ monitor was placed in the target cabin area, and the measured time series was used to characterize the temporal variation of oil mist-dominated particulate pollution during each voyage. For PM₁₀, the comparison dataset included eight untreated voyages, denoted as Tests 1–8, and 13 treated voyages, denoted as Tests 9–21. The PM₁₀ data from each voyage were processed to obtain voyage-level concentration statistics, including average concentration and concentration variability, which were then compared between untreated and treated voyages.

Gas-phase VOCs were monitored by long-term passive sampling in the target cabin areas during each monitored voyage. Tenax TA sampling tubes (60/80 mesh, Camsco, Houston, TX, USA) were used as passive samplers to collect time-integrated VOC samples over the field monitoring period. After sampling, the tubes were sealed, transported to the laboratory, and analyzed using the same GC–MS system (GCMS-QP2010PLUS) as described in Section 2.2.2. The VOC analysis provided TVOC concentrations and major VOC species associated with oil-derived emissions in the ship cabins. For TVOC, the comparison dataset included five untreated voyages and 11 treated voyages. The VOC data were used to evaluate the long-term gas-phase pollution level in the cabins and to compare the field concentration differences between untreated and treated voyages.

The shipboard performance was evaluated by comparing pollutant concentrations between untreated and treated voyages. For a given pollutant i, the apparent field concentration reduction was calculated as Equation (5).

$$R_{\mathrm{i}}={\displaystyle \frac{C_{\mathrm{a}\mathrm{v}\mathrm{g},\mathrm{i},\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d}}-C_{\mathrm{a}\mathrm{v}\mathrm{g},\mathrm{i},\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d}}}{C_{\mathrm{a}\mathrm{v}\mathrm{g},\mathrm{i},\mathrm{u}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d}}}}\times 100\%$$

(5)

where R_i is the apparent field concentration reduction of pollutant i, unitless; C_{avg,i,untreated} and C_{avg,i,treated} are the average concentrations measured during untreated and treated voyages, respectively, expressed in the same concentration unit. In addition to the average concentration level, the concentration range and variability were also compared to assess whether the air-cleaning devices stabilized cabin air quality under fluctuating shipboard operating conditions. The field results therefore represent the practical pollution-control effect of the integrated air cleaners in real ship cabins.

3. Results and Discussion

3.1. Gradient Filter Fabrication and Particle-Removal Performance

The designed gradient filtration structure was fabricated into a full-scale pleated filter module and subsequently evaluated under different airflow rates and inlet PM_2.5 concentrations. As shown in Figure 2a, the fabricated filter retained the intended layered configuration and was assembled within a rigid supporting frame to form a practical filtration component.

Figure 2. Fabrication and particle-filtration performance of the gradient filter module. (a) Fabrication process and prototype of the gradient filter. (b) PM_2.5 single-pass efficiency and initial pressure drop at different airflow rates and face velocities. (c) PM_2.5 filtration performance under different inlet concentrations at an airflow rate of 250 m³/h. (d) Comparison of PM_2.5 removal efficiency and pressure drop with representative HEPA filters reported in the literature, including a fiberglass-based commercial AstroCel HEPA filter (Ref. 1 in panel d) [44], a conventional glass-fiber HEPA filter (Ref. 2 in panel d) [45], and a commercial H13 HEPA filter (Ref. 3 in panel d) [10].

The airflow-dependent filtration performance was first evaluated following the relevant test conditions of the Chinese standard (GB/T 18801-2022), with the upstream PM_2.5 concentration controlled within 0.15–0.75 mg/m³ (Figure 2b). As the airflow rate increased from 150 to 650 m³/h, corresponding to an increase in face velocity from 0.28 to 1.20 m/s, the PM_2.5 single-pass efficiency remained above 99.99% across all tested airflow rates. In contrast, the initial pressure drop increased progressively from 24.5 to 120.8 Pa with increasing airflow rate. At the nominal airflow rate of 250 m³/h, corresponding to a face velocity of 0.46 m/s, the filter achieved a PM_2.5 single-pass efficiency of 99.994% with an initial pressure drop of 41.3 Pa. These results indicate that, within the standard test concentration range, increasing airflow mainly increased the airflow resistance, whereas the particle-removal efficiency remained nearly unchanged.

To further examine the robustness of the filter beyond the standard test concentration range, an independent high-concentration validation was conducted at a fixed airflow rate of 250 m³/h (Figure 2c). When the inlet concentration increased from 0.731 to 11.007 mg/m³, the outlet PM_2.5 concentration remained below 0.001 mg/m³ under all tested conditions, corresponding to single-pass removal efficiencies consistently above 99.99%. Therefore, even when the inlet particle concentration was increased by approximately one order of magnitude beyond the standard test range, the developed filter maintained essentially unchanged PM_2.5 removal performance. Meanwhile, the pressure drop increased moderately from 50.7 to 54.6 Pa with increasing particle concentration. The pressure drop remained below 55 Pa throughout the high-concentration validation, indicating that the filter retained low-resistance operation even under substantially elevated PM_2.5 challenge concentrations. Despite this slight increase in pressure drop, the filter maintained a PM_2.5 removal efficiency above 99.99%. These results demonstrate that the developed gradient filter retained an excellent balance between high filtration efficiency and low airflow resistance under substantially elevated PM_2.5 challenge concentrations.

The efficiency–resistance characteristics of the developed filter were further compared with representative literature-reported HEPA filters (Figure 2d). The present gradient filter achieved a PM_2.5 removal efficiency above 99.99% with a reported pressure drop of 50.7 Pa at a face velocity of 0.46 m/s. At a comparable face velocity of approximately 0.45 m/s, the fiberglass-based commercial AstroCel HEPA filter [44] exhibited a similarly high filtration efficiency but a higher pressure drop, while the conventional glass-fiber HEPA filter [45] showed substantially greater airflow resistance. In comparison, the commercial H13 HEPA filter exhibited a lower pressure drop at a higher face velocity of 0.92 m/s, but its PM_2.5 removal efficiency was clearly lower than that of the present filter. Overall, the comparison demonstrates that the developed gradient filter combines near-complete PM_2.5 removal with relatively low airflow resistance, providing a favorable efficiency–resistance balance compared with the literature-reported HEPA filters.

Taken together, the developed gradient filter maintained PM_2.5 removal efficiencies above 99.99% across both the tested airflow range and the elevated PM_2.5 challenge concentrations up to approximately 11 mg/m³. The consistently high removal efficiency, together with the limited increase in pressure drop under elevated PM_2.5 challenge concentrations, demonstrates the stable high-efficiency and low-resistance performance of the gradient filtration module.

3.2. Activated-Carbon Characterization and Adsorbent Selection

3.2.1. Particle Morphology and Size Distribution

The three candidate activated-carbon adsorbents exhibited distinct particle morphologies and surface textures, as shown in Figure 3. TX-1 showed a relatively regular cylindrical morphology at the particle scale, while its surface was rough and uneven, with evident protrusions and pores. JZT-07 consisted of irregular crushed particles with angular edges and a much more heterogeneous external structure. Its surface was also rough, and the particle shape was clearly less regular than that of TX-1. In contrast, QZ09-4 exhibited a highly regular spherical morphology and the smoothest external appearance among the three adsorbents. At higher magnification, its surface remained comparatively compact and uniform, with much fewer obvious rough protrusions than TX-1 and JZT-07. At the particle scale, these morphology and surface-texture differences primarily affect packing arrangement, bed void fraction, and access of the gas stream to particle surfaces and pore entrances [46,47]. Their influence on dynamic adsorption should therefore be considered together with particle size and the internal pore structure, rather than interpreting external roughness alone as a direct measure of VOC adsorption capacity.

Figure 3. Morphology and particle-size characteristics of the representative activated-carbon adsorbents. (a–c) TX-1; (d–f) JZT-07; (g–i) QZ09-4. Panels (a,d,g) show the particle-scale morphology; panels (b,e,h) show the surface microstructure; panels (c,f,i) show the corresponding particle-size distributions.

The particle-size distributions further reinforced these morphological differences. TX-1 was mainly distributed around 0.9–1.0 mm, indicating a relatively narrow size range for the cylindrical particles. JZT-07 showed a coarser distribution, with most particles larger than 1.0 mm, consistent with its irregular crushed-particle morphology. In contrast, QZ09-4 was mainly distributed within 0.45–0.9 mm and exhibited a comparatively concentrated size range, consistent with its regular spherical morphology. Therefore, JZT-07 was characterized by the coarsest and most irregular particles, TX-1 by rough cylindrical particles of intermediate size, and QZ09-4 by relatively smaller, smoother, and more regular spherical particles.

3.2.2. Pore Structure and Dynamic VOC Adsorption Performance

The bed characteristics, pore structures, and dynamic adsorption performances of TX-1, JZT-07, and QZ09-4 were systematically compared to identify the most suitable adsorbent for the VOC-removal module. To facilitate direct comparison among parameters with different units and magnitudes, the measured bed and pore-structure parameters were normalized, as shown in Figure 4a–c.

Figure 4. Structural screening and dynamic adsorption performance of the representative activated carbons. (a–c) Normalized comparison of bed and pore-structure parameters. (d–f) Single-component breakthrough curves for benzene, toluene, and n-heptane. (g–i) Competitive adsorption behavior in the ternary benzene–toluene–n-heptane system.

Clear differences were first observed in the packed-bed characteristics (Figure 4a). QZ09-4 exhibited the highest packing density of 563.9 g/L and a true density of 2.025 g/cm³, but it had the lowest bed void fraction of 0.28. TX-1 showed intermediate packing characteristics, with a packing density of 475.1 g/L, a true density of 1.987 g/cm³, and a bed void fraction of 0.32. In contrast, JZT-07 exhibited the lowest packing density of 449.0 g/L and a true density of 1.795 g/cm³, while providing the highest bed void fraction of 0.40. Notably, the bed void fraction was calculated from the corresponding packing and particle densities listed in Supplementary Table S1 for details. Therefore, JZT-07 formed the most open packed-bed structure among the three adsorbents, providing a larger fraction of interparticle space for gas transport.

The surface-area characteristics showed an even clearer advantage for JZT-07 (Figure 4b). Its BET specific surface area reached 1233 m²/g, compared with 1002 m²/g for TX-1 and 826 m²/g for QZ09-4. The same trend was observed for the micropore surface area, which was 1136, 871, and 752 m²/g for JZT-07, TX-1, and QZ09-4, respectively. JZT-07 therefore exhibited both the largest overall surface area and the most developed microporous surface among the three adsorbents.

The pore-volume comparison further confirmed this structural advantage (Figure 4c). JZT-07 exhibited the highest total pore volume of 0.59 cm³/g, followed closely by QZ09-4 at 0.57 cm³/g and TX-1 at 0.51 cm³/g. A more pronounced difference was observed in micropore volume. JZT-07 reached 0.47 cm³/g, compared with 0.38 cm³/g for TX-1 and 0.31 cm³/g for QZ09-4. Thus, JZT-07 combined the highest bed void fraction with the largest BET surface area, micropore surface area, total pore volume, and micropore volume, giving it a highly developed pore structure together with a relatively open packed-bed configuration among the tested adsorbents. Accordingly, the favorable dynamic performance of JZT-07 was associated with the combined effects of its relatively open interparticle structure, which supported external gas transport and contact, and its developed microporous texture, which provided the principal adsorption space.

The single-component breakthrough tests were subsequently used to evaluate whether these structural advantages translated into improved dynamic VOC adsorption at an inlet concentration of 100 ppm for each tested compound (Figure 4d–f). For benzene, the breakthrough curve of JZT-07 was clearly shifted toward longer adsorption times compared with TX-1 and QZ09-4. JZT-07 achieved a breakthrough capacity of 132 mg/g and a saturation capacity of 167 mg/g, compared with 100 and 120 mg/g for TX-1 and 86 and 102 mg/g for QZ09-4, respectively. A similar trend was observed for toluene, for which JZT-07 reached breakthrough and saturation capacities of 167 and 207 mg/g, higher than those of TX-1 and QZ09-4. For n-heptane, JZT-07 again exhibited the strongest dynamic adsorption performance, with breakthrough and saturation capacities of 159 and 197 mg/g. Therefore, JZT-07 consistently showed the latest breakthrough and the highest adsorption capacities for all three representative VOCs. In addition to the differences among adsorbents, clear differences were also observed among the tested VOCs. For all three activated carbons, the saturation capacity for benzene was lower than those for toluene and n-heptane. This trend can be partly attributed to the lower molecular weight of benzene. In comparison, the higher molecular weight and stronger dispersive intermolecular interactions of toluene and n-heptane can enhance their affinity toward carbonaceous pore walls, demonstrating the key factor of intermolecular forces in adsorption [48].

The adsorption behavior became more complex when benzene, toluene, and n-heptane were simultaneously introduced at an inlet concentration of 100 ppm for each component, corresponding to a total VOC concentration of 300 ppm (Figure 4g–i). For all three adsorbents, benzene broke through first, followed by n-heptane and then toluene, indicating clear competitive adsorption among the three VOCs. A pronounced concentration overshoot was observed for benzene. After the initial adsorption stage, the outlet benzene concentration increased above its inlet concentration of 100 ppm and subsequently decreased toward the inlet level. This behavior indicates that benzene adsorbed during the early stage was progressively displaced from adsorption sites by the more strongly retained competing VOCs [49]. The released benzene was superimposed on the continuously incoming benzene stream, resulting in a transient outlet concentration exceeding the inlet concentration. As the displacement process gradually weakened, the outlet benzene concentration decreased toward the inlet level.

The quantitative breakthrough and saturation parameters obtained from the ternary VOC tests are summarized in Table 1. Clear differences were observed among the three adsorbents under competitive conditions. For benzene, QZ09-4 showed the highest breakthrough capacity of 48.0 mg/g, followed by JZT-07 at 45.8 mg/g and TX-1 at 38.9 mg/g. A similar trend was observed for toluene, with breakthrough capacities of 86.4, 83.8, and 81.0 mg/g for QZ09-4, JZT-07, and TX-1, respectively. In contrast, JZT-07 exhibited the highest n-heptane breakthrough capacity of 93.8 mg/g, compared with 88.1 mg/g for QZ09-4 and 82.2 mg/g for TX-1.

Table 1. Breakthrough and saturation adsorption parameters of the three adsorbents in the ternary VOC system.

Adsorbent	VOC Component	Breakthrough Time, tb (min)	Breakthrough Adsorption Capacity, qVOC,b (mg/g)	Saturation Time, ts (min)	Saturation Adsorption Capacity, qVOC,s (mg/g)
JZT-07	Benzene	120	45.8	186	61.8
	Toluene	186	83.8	450	142.4
	n-Heptane	192	93.8	390	142.8
TX-1	Benzene	102	38.9	162	52.3
	Toluene	180	81.0	312	112.8
	n-Heptane	168	82.2	282	115.9
QZ09-4	Benzene	126	48.0	186	64.0
	Toluene	192	86.4	462	145.0
	n-Heptane	180	88.1	300	118.0

Note: Breakthrough and saturation were defined as C_VOC,out/C_VOC,0 = 0.05 and 0.95, respectively. q_VOC denotes the dynamic adsorption capacity calculated from the integrated inlet–outlet VOC concentration difference.

The saturation capacities showed that JZT-07 maintained a strong and balanced adsorption performance for all three VOCs. Its saturation capacities reached 61.8, 142.4, and 142.8 mg/g for benzene, toluene, and n-heptane, respectively. QZ09-4 showed comparable capacities for benzene and toluene, but its n-heptane saturation capacity was lower at 118.0 mg/g, while TX-1 exhibited lower overall capacities under the ternary condition. These results demonstrate that JZT-07 retained substantial adsorption capacity across aromatic and aliphatic VOCs even in the presence of strong competitive adsorption. Taken together, the structural screening, single-component breakthrough tests, and ternary competitive adsorption results supported the selection of JZT-07 as the final adsorbent for engineering integration. Its open packed-bed structure and developed microporosity provided favorable structural characteristics, while its dynamic adsorption performance remained high across all three representative VOCs. Although QZ09-4 showed slightly higher benzene and toluene capacities in the ternary system, JZT-07 combined the strongest single-component adsorption performance, the highest n-heptane capacity under ternary exposure, and a balanced multi-component adsorption profile. JZT-07 was therefore selected as the final adsorbent for the VOC-removal module of both integrated air cleaners.

3.3. Standard-Chamber Performance of the Integrated Air Cleaners

Following the component-level optimization of the gradient filter and activated-carbon adsorbent, the selected filtration and adsorption modules were integrated into the Type A accommodation-cabin and Type B machinery-cabin air cleaners. Their device-level performance was subsequently evaluated in the standard test chamber, and the principal performance parameters are summarized in Table 2.

Table 2. Device-level performance of the Type A and Type B air cleaners.

Parameter	Type A Accommodation-Cabin Air Cleaner	Type B Machinery-Cabin Air Cleaner
Rated power (W)	32.8	158
Airflow rate (m3/h)	78.0	413
PM2.5 CADR (m3/h)	-	369.68
Xylene CADR (m3/h)	64.39	308.14
PM2.5 purification energy efficiency (m3/(h·W))	-	2.34
Xylene purification energy efficiency (m3/(h·W))	1.96	1.95
Operating noise (dB(A))	46.0	55.0
Dimensions, L × W × H (mm)	330 × 300 × 215	680 × 340 × 415
Mass (kg)	10.0	30.0

Note: “-” indicates that the Type A PM_2.5 CADR result was excluded following a data-quality audit because sufficiently consistent and comparable chamber-test conditions could not be verified from the archived records.

The two air cleaners exhibited clearly different airflow and treatment capacities in accordance with their intended application scales. The Type A air cleaner operated at an airflow rate of 78 m³/h with a rated power of 32.8 W, whereas the Type B air cleaner delivered 413 m³/h at 158 W. For the retained standard-chamber datasets, the Type A xylene CADR was 64.39 m³/h, while the Type B PM_2.5 and xylene CADRs were 369.68 and 308.14 m³/h, respectively. The Type A PM_2.5 CADR result was excluded from the quantitative analysis following the data-quality audit described in Section 2.2.3. Thus, the higher-capacity Type B configuration provided substantially greater particle- and VOC-treatment capacity, while the compact Type A configuration retained effective xylene clean-air delivery for accommodation-cabin use.

The corresponding purification energy efficiencies were 1.96 m³/(h·W) for Type A xylene and 2.34 and 1.95 m³/(h·W) for Type B PM_2.5 and xylene, respectively. These results indicate that the higher-capacity Type B configuration delivered substantially greater clean-air output while maintaining favorable pollutant-removal performance relative to its power demand.

The increased treatment capacity of the Type B unit was accompanied by a larger device size and a moderate increase in operating noise. The Type A air cleaner had dimensions of 330 × 300 × 215 mm and a mass of 10.0 kg, with an operating noise of 46 dB(A). The Type B air cleaner was scaled to 680 × 340 × 415 mm and 30.0 kg, while the operating noise increased to 55 dB(A). Therefore, the Type A configuration provided a compact and relatively low-noise solution for accommodation cabins, whereas the Type B configuration emphasized substantially higher airflow and purification capacity for machinery cabins with stronger pollutant loads. Overall, the device-level results demonstrate that the same two-stage filtration–adsorption strategy could be implemented at different scales. The Type A air cleaner provided compact and low-noise operation with effective xylene clean-air delivery for accommodation spaces, while the Type B air cleaner delivered substantially greater PM_2.5 and xylene clean-air capacity for high-pollution machinery cabins.

3.4. Shipboard Field Validation

The shipboard performance of the integrated air cleaners was evaluated by comparing PM₁₀ and TVOC concentrations during untreated and treated voyages in accommodation and machinery cabins, as shown in Figure 5. Clear reductions in both particulate and gaseous pollutant concentrations were observed during treated voyages, demonstrating that the filtration–adsorption air cleaners maintained effective pollutant control under actual shipboard operating conditions. The two cabin types also exhibited distinct baseline pollution characteristics, which provided a practical basis for evaluating the compact Type A and higher-capacity Type B configurations in their intended application environments.

Figure 5. Multi-voyage shipboard performance of the integrated air cleaners. (a) PM₁₀ concentrations in the accommodation cabin. (b) TVOC concentrations in the accommodation cabin. (c) PM₁₀ concentrations in the machinery cabin. (d) TVOC concentrations in the machinery cabin. Untreated and treated voyages represent shipboard operation without and with the corresponding air cleaners, respectively.

In the accommodation cabin, the average PM₁₀ concentration decreased from 28.0 μg/m³ during untreated voyages to 6.92 μg/m³ during treated voyages (Figure 5a), corresponding to a reduction of approximately 75.3%. The PM₁₀ concentrations during the untreated voyages also showed substantial voyage-to-voyage variation, whereas the treated-voyage concentrations remained consistently at a much lower level. A similar improvement was observed for gas-phase pollutants (Figure 5b). The average TVOC concentration decreased from 5834 to 1962 μg/m³, corresponding to a reduction of approximately 66.4%. The consistently lower PM₁₀ and TVOC concentrations during treated voyages demonstrate that the Type A air cleaner provided simultaneous control of particle- and gas-phase pollutants in the accommodation environment.

The machinery cabin exhibited substantially higher particulate pollution levels, with an average untreated PM₁₀ concentration of 202 μg/m³ (Figure 5c). After deployment of the Type B air cleaner, the average concentration decreased to 69.0 μg/m³, representing a reduction of approximately 65.8%. More importantly, the large PM₁₀ concentrations observed during several untreated voyages were no longer present during treated operation, and the treated-voyage concentrations were concentrated within a substantially lower range. The same trend was observed for TVOC (Figure 5d). The average concentration decreased from 4349 μg/m³ during untreated voyages to 1322 μg/m³ during treated voyages, corresponding to a reduction of approximately 69.6%. These results demonstrate that the higher-capacity Type B air cleaner maintained substantial particulate and VOC removal even under the more heavily polluted machinery-cabin conditions.

The multi-voyage results further demonstrate the complementary roles of the two air-cleaner configurations. Both configurations consistently reduced particulate and gaseous pollutant levels despite the clear differences in pollution characteristics and source intensities between the two cabin environments. The recurring lower concentrations across multiple treated voyages show that the component- and chamber-level performance was successfully translated into practical shipboard pollutant reduction.

3.5. Limitations and Future Work

There are still several aspects of the present study that could be further investigated. The activated-carbon module was operated as a replaceable packed bed, and the present study mainly evaluated its dynamic adsorption performance through breakthrough experiments and its practical VOC-control performance after integration into the air-cleaning devices. Adsorption–regeneration cycling was not included in the present experimental program. The regeneration stability and adsorption-capacity retention of JZT-07 after repeated use remain valuable topics for further investigation, together with the optimization of thermal, vacuum, or purge-gas regeneration conditions.

The characterization of the commercial activated carbons in this study focused on particle morphology, packed-bed characteristics, pore structure, and dynamic VOC adsorption, because these parameters were directly associated with the adsorption and transport behavior considered in the device development. More detailed information on carbon structural ordering was not obtained in the present study. X-ray diffraction and Raman spectroscopy could therefore be introduced in further work to characterize the carbon structure and examine possible structural changes after prolonged use or regeneration.

The shipboard tests were conducted under the routine operating and maintenance conditions of the air cleaners. During navigation, the devices were operated continuously and the cartridges were replaced every 15 days. Periodic monitoring showed that the daily mean VOC and particulate concentrations remained generally stable, although short-term variations occurred with changing shipboard operating conditions. The present field results therefore characterize the practical pollutant-control performance under this maintenance regime rather than the continuous aging process of an individual activated-carbon bed. Further field measurements following the same carbon bed throughout a complete service period, together with cumulative operating time and VOC loading, would allow a more quantitative assessment of adsorption saturation and replacement criteria under actual shipboard conditions.

Because the archived Type A PM_2.5 chamber-test records did not permit verification of sufficiently consistent and comparable test conditions, this dataset was excluded from the revised quantitative analysis. Future work should re-evaluate this performance metric using a unified standard-chamber protocol with verified device airflow, chamber mixing, consistent fitting intervals, and explicit uncertainty analysis.

4. Conclusions

This study developed and evaluated an integrated filtration–adsorption air-cleaning strategy for simultaneous control of oil mist-dominated particulate matter and VOCs in shipboard cabins. The gradient filtration module maintained PM_2.5 single-pass efficiencies above 99.99% over airflow rates of 150–650 m³/h. At the nominal airflow rate of 250 m³/h, the removal efficiency reached 99.994% with an initial pressure drop of 41.3 Pa. When the inlet PM_2.5 concentration was further increased to approximately 11 mg/m³, the outlet concentration remained below 0.001 mg/m³ and the removal efficiency remained above 99.99%, while the pressure drop stayed below 55 Pa. These results demonstrate stable high-efficiency and low-resistance filtration performance across airflow variation and elevated PM_2.5 challenge conditions.

Among the activated carbons, JZT-07 provided the strongest overall balance of packed-bed structure, pore characteristics, and dynamic adsorption performance. It exhibited the highest bed void fraction (0.40), BET specific surface area (1233 m²/g), micropore surface area (1136 m²/g), total pore volume (0.59 cm³/g), and micropore volume (0.47 cm³/g) among the three candidates. Its single-component breakthrough capacities reached 132, 167, and 159 mg/g for benzene, toluene, and n-heptane, respectively. Under ternary exposure, clear competitive adsorption and benzene concentration overshoot were observed, reflecting displacement of previously adsorbed benzene by more strongly retained VOCs. Considering the structural characterization, single-component adsorption capacity, and balanced ternary adsorption profile, JZT-07 was selected for the VOC-removal module.

The selected filter and JZT-07 packed bed were integrated into two air-cleaner configurations for different cabin demands. In the standard chamber, the Type A accommodation-cabin air cleaner delivered 78 m³/h with a xylene CADR of 64.39 m³/h, while the Type B machinery-cabin unit delivered 413 m³/h with PM_2.5 and xylene CADRs of 369.68 and 308.14 m³/h, respectively. The corresponding purification energy efficiencies were 1.96 m³/(h·W) for Type A xylene and 2.34 and 1.95 m³/(h·W) for Type B PM_2.5 and xylene, respectively. The Type A configuration emphasized compact and low-noise operation, whereas the Type B configuration provided substantially greater clean-air capacity for more heavily polluted machinery spaces.

The multi-voyage field evaluation confirmed that the controlled chamber performance translated into practical shipboard pollutant reduction. In accommodation cabins, the average PM₁₀ and TVOC concentrations decreased by 75.3% and 66.4%, respectively, during treated voyages. In machinery cabins, the corresponding reductions were 65.8% and 69.6%. The recurring lower pollutant concentrations across multiple voyages demonstrate that the two-stage filtration–adsorption route can provide simultaneous particle and VOC control under real shipboard operating conditions. Overall, the study establishes a coherent development pathway from component-level design and adsorbent screening to device-scale verification and field application in confined occupied environments.