Determination of performance-parameter design and impact factors of sampling efficiency for bioaerosol cyclones

Abstract

Bioaerosol sampling is a critical step toward characterizing airborne pathogen exposure risks for many respiratory infectious diseases in human-to-human or human-to-animal outbreaks. The integrated performance-parameter design and sampling efficiency of aerosol-to-hydrosol samplers (cyclones) has been rarely published or studied. This review focuses on the determination of cyclone performance-parameter design and impact factors of the sampling efficiency, discusses their mechanisms of collection and describes their advantages and drawbacks. The crucial design parameters of cyclones in different applications are then discussed in detail. We also discuss how the aerosol-to-hydrosol sampling efficiency is intrinsically related to the performance parameters for bioaerosol cyclone design. In the final section of this review, scientific questions and challenges for future work are presented to encourage high collection efficiency design and high-quality cyclones.

Keywords:

• Bioaerosol
• cyclone
• biosampler
• sampling efficiency
• aerosol-to-hydrosol

Introduction

Bioaerosols are suspensions of airborne biological materials, including bacteria, viruses, proteins, spores, pollen, fungi, algae and cell fragments. They range in size from 0.01 to 100 µm [1]. Pathogenic bioaerosols are a key reason that many emerging or re-emerging respiratory infectious diseases occur in human-to-human or human-to-animal outbreaks. For example, the viruses causing severe acute respiratory syndrome (SARS), H1N1 influenza, Middle East respiratory syndrome (MERS) and the novel coronavirus disease (COVID-19) can transmit through aerosols and cause thousands of deaths [2–6]. Some transmissions through bioaerosols are fast and invisible, for example, there were at least 742 confirmed COVID-19 cases on the Diamond Princess cruise ship containing 3,700 people as the result of a superspreading event [7]. Highly efficient bioaerosol samplers are necessary to capture these airborne biological particles with better viability to perform further biological analysis. Aerosol-to-hydrosol cyclones usually have high flow rates to collect large quantities of bioaerosols, after which they collect the bioaerosol as a concentrated liquid to create a cushioning effect and prevent violent impacts between the aerosol and collection media [8,9]. For monitoring a healthcare environment, an ideal cyclone should have the following characteristics: a high sampling efficiency, a fast and continuous sampling function, a high particle deposition efficiency, a proper particle size selection, a wide range of particle cut-off diameters, and a good microorganism viability [1,8,10]. However, there are many performance parameters for cyclone design that affect the sampling efficiency of aerosol-to-hydrosol cyclones [8]. The performance-parameter design and the sampling efficiency of aerosol-to-hydrosol cyclones have been rarely studied together. This review is focused on the performance-parameter design and the factors that affect the sampling efficiencies of bioaerosol cyclones. The collection mechanisms and the advantages and drawbacks of bioaerosol cyclones are described, and the crucial design parameters of cyclones in different applications are discussed in detail. Finally, this review addresses the design requirement for cyclones to achieve a high sampling efficiency.

Bioaerosol sampling

Category of current bioaerosol samplers

Bioaerosol sampling research has developed for over 200 years. Since their first introduction by Maddox in the 1800s [11], bioaerosol samplers have advanced as the understanding of the aerodynamics and particle collection, and have improved and developed computational fluid dynamics methods [1]. Macher [12] stated that there are three key points related to the sampling performance: the inlet sampling efficiency for bioaerosol particles aspirated from the environment, the particle-capturing efficiency at the filter or laboratory culture, and the biological recovery efficiency for the viability of bioaerosol particles [1]. From the 1900s to present, a large number of commercial bioaerosol sampling designs have been developed for different purposes, as summarized in Table 1.

Table 1. Categories of bioaerosol samplers.

Samplers	Collection mechanism	Pros	Cons	Air flow rate	Particle size range (aerodynamic diameter)	References
Nutrient plates	Passive inertial	Easy-to-use low-cost, no need for post-sampling preparation	Long culturing time (48 h), no sampling quantitative measurements for microorganisms and air volume	Natural	Any	[1] [8] [13]
Surface sampler	Passive inertial	Easy-to-use, low-cost, no need for post-sampling preparation	Long natural culturing time (48 h), no sampling quantitative measurements for microorganisms and air volume	Natural	Any	[1] [8] [13]
Six stage viable cascade impactor	Active inertial	Particle size separation	Better for lower-concentration bioaerosols, damages microorganisms with violent impact	28.3 L/min	0.65–7 μm	[1] [14]
Impinger	Active inertial	Better pathogen viability, higher collecting efficiency for cut-off size of 1–3 μm	Low flow rate, liquid evaporation	12.5 L/min	0.3 μm	[14] [15]
Wetted-wall cyclone	Active inertial	High flow rate, better pathogen viability, good collection efficiency	Loses smaller and lighter particles, liquid evaporation,	60–400 L/min	0.5–1 µm	[15] [16]
Batch-type wetted wall cyclone	Active inertial	High flow rate, better pathogen viability, no liquid evaporation, good collection efficiency	Low collecting efficiency for cut-off size ≤ 1 µm, and ≥ 10 µm, losses of large particles in the inlet nozzle	265–307 L/min	1–10 µm	[17] [18] [19]
Virtual cyclone	Active inertial	Better pathogen viability, better collection efficiency for small cut-off size particles, good collection efficiency	Unclear relationship between inlet/outlet width and the collection efficiency, unclear separation mechanism	10–50 L/min	0.8–3.2 μm	[20] [21] [22]
Filters	Inertial force, diffusion, and electrostatic attraction	Simple, low cost, and effective	Small filter size and limited flow rate, better for lower-concentration bioaerosols	4 L/min	0.2–3 μm	[5] [23] [24]
Electrostatic precipitation	Others	More concentrated culturable samples for bacteria	Nonviable, lower concentrations of fungal spores	1–10 L/min	0.2–3 μm	[25] [26] [27]

Inertial samplers use gravity and external air flow to deposit biological particles onto a collection surface. These samplers can be either passive or active. Passive samplers have the advantage of not disturbing the surrounding air flow but losing smaller and lighter particles [28,29]. Active samplers include impactors, impingers and cyclones. For these samplers, smaller and lower-mass particles easily pass through a bent pipe and are collected by the medium, but larger and heavier particles hit the wall of the pipe and hence cannot traverse the bend due to inertial compaction. Furthermore, the impinger has a drawback in terms of collection losses from the evaporation of the collection liquid and the adhesion of the particles to the collection container wall [26].

Sampling by filtration, in which the bioaerosol particles are trapped on fine fibres or porous membrane surfaces, is another way to collect particles. However, filters are less popular for bioaerosol particle collection because the bio-efficiency is low due to the desiccation of the pathogens and the adherence to the filter fibres [28].

Cyclones

In recent applications, cyclones have been widely used as highly efficient samplers. King and McFarland [15] showed that the culturability of Escherichia coli collected with a cyclone was about a factor of 100 higher than that collected with an impactor. They also found that real-time polymerase chain reaction (RT-PCR) showed E. coli identification was 100 times more concentrated by cyclone collection rather than impactor collection [15]. Cyclones feature large air-flow rates, concentrated-liquid sampling processes, and better viabilities for biological particle collection [17,30].

In most cyclones, the bioaerosol enters the inlet and experiences a centrifugal force based on the shape of a spiral collection tube and the swirling air flow [1,8,20,31], as shown in Figure 1. Within this airflow, the particles are acted upon by a centrifugal force proportional to their diameters, densities and speeds. McFarland et al. [32] separated cyclones into two groups. They are wetted wall cyclones (WWCs) and batch wetted wall cyclones (BWWCs). WWCs use the centrifugal force to drive particles under inertia toward the liquid containers from an aerosol to a hydrosol [28]. Cage et al. [33] designed a WWC with a flow rate of 400 L/min and concentrated the sample in 12 mL of liquid. Another WWC, designed by Sigaev et al. [34], had a flow rate of 60 L/min and concentrated the sample in 10 mL of hydrosol with a cut-off size of the collected particles of 0.5–1 µm (aerodynamic diameter, AD). Commercial WWCs include the SASS air collectors produced by Research International Inc. (Monroe, WA) with 265 L/min flow rates that concentrate the sample into 5 mL of liquid and the air samplers of Zaromb Research Corp (Burr Ridge, IL) with 300 L/min flow rates that concentrate the samples into 25 mL of liquid. Buchanan et al. [35] developed an improved version of the WWC with a sampling flow rate of 950 L/min and a particle cut-off size of less than 1 µm (aerodynamic diameter, AD). Black & Shaw [36] applied the liquid through a port at the air inlet slot and added a vortex finder at the cyclone body. The air flow rate was 900 L/min, and the pressure drop across the cyclone was 7 kPa. The collection efficiency was about 60% for bioaerosol particles in the AD size range of 1–7 µm. A phosphate buffered saline solution with Tween-20 surfactant was used as a collection buffer [36]. Based on the design of Black and Shaw [36], the WWC-1250 device was detailed by McFarland et al. [32] with four modifications (Figure 2). As reviewed in King et al. [13], in addition to continuous WWCs, there are batch wetted wall cyclones (BWWC), where the particles are collected in liquid batches that are replaced on pre-set time intervals. Similar to the WWCs, BWWC systems are advantageous because they support long sampling periods, which allows for more concentrated samples and avoids the evaporation problems of impinging samplers [18,19].

Figure 1. Cyclone with swirling spiral pattern (Reused from Peng et al. [31] with permission).

Figure 2. WWC-1250 device with four modifications by McFarland et al. [32] [Reused with permission]. Red circled numbers: $1$ the aerosol inlet of the upgraded device forms a converging flow path into the cyclone, $2$ the liquid is atomized before entering the inlet slot of this device, $3$ the length of the cylindrically shaped cyclone body is shorter, and $4$ the skimmer was changed to a straight shape rather than a divergent opening.

Another type of cyclone is the virtual cyclone, which has a collecting probe operating with a major and a minor flow. Particles enter the cyclone and are carried by the airflow around a bend. By moving with the air, the smaller particles navigate the curve and are deposited on a collection filter. In contrast, the larger particles having greater inertia are carried into a different collection container (Figure 3).

Figure 3. Virtual cyclone. Left: airflow in the virtual sampler (Reused from Haig et al. [28] with permission). Right: schematic of a virtual cyclone (Reused from Song et al. [20] with permission).

Currently, a variety of conventional cyclones have been constructed and designed for different bioaerosol sampling purposes and working environments. The purpose of these designs is to achieve cyclones with high sampling efficiencies and to integrate them with detection systems for consistent sample delivery, automated operation and portability. Widely used commercial cyclones (Figure 4) include liquid-medium samplers such as the Coriolis®µ Biological Air Sampler with a 300 L/min flow rate (Bertin Technologies, Saint Quentin en Yvelines, France), the SASS 2300 with a 325 L/min flow rate (Research International, Inc., Monroe, WA, USA), the BioCapture650 with a 200 L/min flow rate (FLIR Systems, Inc. Wilsonville, OR, USA), the large-particle collecting cyclone S110 with a 10 L/min flow rate (Dekati, Kangasala, Finland), the Burkard cyclone sampler with a 16.6 L/min flow rate (Burkard manufacturing Co. Ltd, UK) and the BIO-Capturer-6 with an 80 L/min flow rate (Fuji Biotechnology, Inc., Hangzhou, China). Some cyclones perform particle-size selection before the air enters the inlet, and the particle-size selection head has five axial flow stages from 0.32 to 5.4 µm at a sampling flow rate of 28.3 L/min [37]. Hsiao et al. [38] developed a particle-size selection cyclone with a cut-off particle diameter from 0.04 to 11.3 µm. Tan et al. [39] described the development of an automated electrostatic sampler that collected particles in a liquid reservoir and delivered them to sensing devices. Other reports have described a direct analysis system of airborne pathogens based on microfluidic enrichment [40–42]. Another device, the MicroSampler, which is based on a two-phase fluid control in a microchip, provides real-time bioaerosol sensing [43]. A recent study has focused on improvements in the real-time and automated operations of cyclones [10]. However, few fully integrated bioaerosol sampling systems are capable of continuous real-time sampling. All of these cyclones have large particle cut-off diameters and low particle collection and aerosol-to-liquid efficiencies [15, 44].

Figure 4. Recent commercial bioaerosol cyclones: Coriolis®µ (a), Research International SASS2300 (b), FLIR Systems BioCapture650 (c), Dekati S110 (d), Burkard C90M (e) and Fuji Biotechnology BIO-Capturer-6 (f).

In most cyclones, larger particles are easier to collect than smaller particles. The air carries small particles at the bottom of the collection container, allowing them to escape out of the cyclone at the top of the outlet [28]. Sampling losses may again arise from the evaporation of the collection liquid and the adherence to the wall of the collection container. The capturing efficiency of biological particles is affected by both the airflow rate and the cyclone geometry, including the size of the airflow inlet, the rate of the inlet airflow, and the geometric shape of the collection container. Apart from new mechanical geometric designs, the optimization of the cyclone’s sampling performance parameters are the most significant factors for achieving high-sampling-efficiency bioaerosol cyclones.

Crucial design and performance parameters of cyclones

Performance parameters

In the evaluation of the cyclone collection performance, the bioaerosol capturing efficiency is indispensable. There are no standards for evaluating the integrated performances for cyclones [45]. However, there are three measurements used to describe the aerosol collection efficiency (Figure 5). Most studies presumed that higher values of these three efficiencies can be used to evaluate the total sampling efficiency for cyclones [1,8]. First, the inlet efficiency (aerosol-to-aerosol efficiency, η_AA) is related to the ratio of the concentration of particles at the inlet entrance to that from the sampled aerosol. Second, the transport efficiency (aerosol-to-hydrosol efficiency, η_AH) is the ratio of the product of the particle concentration in a liquid or collection medium and the liquid flow rate to the product of the particle concentration and the air flow rate at the inlet. Finally, the deposition efficiency is the ratio of the concentration of particles in the liquid to the concentration of particles at the inlet entrance [32]. In most studies on cyclone performance, improvements and developments were achieved through either simulations or experiments, but not through their integration [46].

Figure 5. Venn diagram of the three cyclone performance parameters associated with the inlet efficiency, transport efficiency, and deposition efficiency. (1) Yellow: the common factors associated with all parameters. These are the sampled air-flow rate, particle concentration in the air stream, and particle size. (2) Blue: common factors between the inlet and deposition efficiencies. These are the particle concentrations in the air at the inlet. (3) Pink: common factors between the sampling and deposition efficiencies. These are particle concentrations in the liquid medium. (4) Green: common factors between the inlet and transport efficiencies. These are sampled with the air flow rate [8].

Experimental method

Generally, the air flow in a cyclone is too complex and turbulent to calibrate accurately in a laboratory set-up. Moreover, there are no international standards for testing the sampling efficiencies of cyclone bio-samplers or measuring the flow structures. Then, determining effective geometries of the devices is difficult. Many previous studies used wet-lab experimental data to evaluate the cyclone performance based on these three ratios [17,20,47–50]. In some experiments, inactive microorganism simulants, such as polystyrene latex (PSL), were applied to test the aerosol-to-hydrosol collection efficiencies of cyclones [15,51,52]. McFarland et al. [32] defined the aerosol-to-hydrosol efficiency to be the intrinsic system of the WWC performance parameter with the detection of polystyrene latex spheres. They showed an aerodynamic diameter (AD) cut-off of 1.2 µm and an average collection efficiency of 90% over the AD size range of 2–10.2 µm. In addition, the cyclone performance evaluation with biological particles is a critical performance characterization. A previous study showed that the collection (aerosol-to-hydrosol) efficiency of a cyclonic swirling bioaerosol sampler for an aerosol cut-off size larger than 0.7 μm was 50%, and for that larger than 2 μm was 80% [52]. In another case, the sampling efficiency was tested for two new cyclone-based bioaerosol samplers. PSL particles ranging from 0.5 to 10 μm were used as the test aerosols for acquiring sampling efficiencies. They found that the capture (aerosol-to-hydrosol) efficiency was 70%–80% for aerosol particles larger than 5 μm [53]. In another study of a designed BWWC, the collection efficiency was 51% for 1 µm PSL and 55% for 3 µm PSL [18].

Although the above results were not the same, the overall analytical method for determining the sampling efficiency was similar. Whether calling it sampling or collection or capturing efficiency, all of the previous studies examined the single aerosol-to-hydrosol efficiencies of cyclones based on one or two factors. No integrated evaluations for aerosol-to-aerosol, aerosol-to-hydrosol and deposition efficiencies have been performed. Time-consuming and expensive wet-lab experiments were performed using thousands of factor combinations, including the particle size, flow rate, pressure drop and movement dynamics, which were related to the sampling efficiency to determine the optimal combination by trial-and-error approaches. Additionally, the one-by-one testing of each parameter in a biological lab may be prone to evaluation biases, leading to a failure to obtain comprehensive results and to compare the sampling efficiencies between different bio-sampler designs. Hence, more comprehensive considerations using theoretical models are needed to evaluate many different influence factors.

Numerical design

Numerical simulations are useful means to study cyclones having a large number of influence factors and to optimize the integrated set of parameters to attain a high sampling efficiency and a better geometric design [54]. Good predictive models, whether empirical, semi-empirical, or mathematical, can be used to design new cyclones for new applications or to improve existing designs [30]. As reviewed by Zhao [54], historically, Beekmans & Kim [55] developed a quadratic response surface model (RSM) to evaluate the cyclone efficiency. In their model, the Reynolds number (R_e) and Stokes number (S_tk) were included, but the effects of the cyclone geometric parameters were not involved. In later studies, the Stokes and Reynolds numbers are two indicators that must be evaluated in the design of geometrically similar cyclones. They are related to the density of the liquid, aerodynamic particle diameter, viscosity, air flow rate at the inlet, air density, and inlet slot width, which are all related to sampling efficiency [8,54].

Computational fluid dynamics (CFD) has become a widely accepted design tool for research and development in fluid mechanics over the last decade. It can be used to optimize cyclone performance through modifications to the geometric design and different operational performance parameters [56–59]. CFD models provide more details of the flow structures and particle behaviors related to the sampling efficiency, and they are crucial factors for guiding the geometric design parameters, including the sampling inlet, transport, and deposition of a cyclone. For example, Hu and McFarland [16] redesigned the geometry and increased the flow rate of an earlier cyclone [36] (Figure 6). Previous CFD evaluations of the sampling efficiency were based on several single point measurements for each parameter using trial-and-error approaches [60–62]. Many CFD models have been used and optimized to explore the parameter space of cyclones, which includes the flow rates, collection characteristics and pressure drop, liquid film thickness and width, liquid medium, and material for the liquid collector. All of these parameters were consistent with the experimental results [63, 57,64,65]. For example, Wang et al. [62] focused on reducing the pressure drop by adding a stick to the flow field of a cyclone. Another simulation model showed the strong effect of the geometric swirl number on the mean flow characteristics for cyclones [61].

Figure 6. Modifications in aerosol-to-hydrosol cyclone based on CFD simulations (Reused from Hu & McFarland [16] with permission): $1$ narrowing the rectangular entrance slot, $2$ narrowing the shape of air-flow inlet, $3$ adding water injection at the top of cyclone inlet, $4$ changing to a rectangular cyclone body and $5$ re-designing the shape of the cyclone skimmer in the rectangular region.

In the previous CFD simulation examples listed above, each point measurement corresponded to a single factor or parameter in a numerical model, and wet-lab experiments used an exhaustive method to evaluate the optimal parameter values of a cyclone. However, most geometric designs, aerosol flow structures, particle concentrations, and movement dynamics are correlated to each other. Many factors in the theoretical formulas must be considered and optimized in an integrated system under various restricted conditions. In some previous studies, empirical models were used based on the understanding of the flow structure and the experimental data in the cyclone to predict the particle dynamics and flow structure and to analyse the design parameters of the cyclone to develop cyclone modifications that promote higher collection efficiencies [66–68]. These models based on finite-element methods are able to calculate the particle collection efficiency, cut-point particle size, pressure drop, and different geometries of cyclones.

Although several numerical simulations have been performed by CFD, they have some limitations, including the modeling reliability, the CPU processing ability, and mesh generation difficulties. The actual systems being simulated were usually complex, and the CFD simulations were often simplified, such as using simplified physical models, mathematical models, and boundary conditions, which may have affected the reliability of the simulations [69]. The quality of the CFD numerical solutions strongly depends on the type of turbulence model applied to the continuous phase flow [70]. This is another possible uncertainty affecting the calculating accuracy arising from the choice of the mathematical model. Currently, CFD can achieve many functions, and some software provides user-defined function assistance. However, there are still some complex cases that cannot be modeled, and programming is needed for simulation accuracy [46]. Most CFD models include a large number of factors. Such complicated simulations require long CPU processing times and may result in data divergence [71]. In addition, CFD simulations require more effort in the pre-processing for mesh generation. This is especially significant for simulating a complex cyclone model. Inadequate mesh generation, such as improperly selected mesh ratios and sizes, may result in uncertainties of the CFD modeling or divergent results [72]. Recently, a meshless approach based on the LBM (Lattice‒Boltzmann Method) was employed for simulations. The concept of discrete grids was replaced by a lattice in the LBM. Furthermore, instead of solving Navier–Stokes equations like other CFD methods, this method calculates the streaming and collisions to simulate the motion of the whole fluid and predicts the hydrodynamics by tracking the probability distribution function to an equilibrium state. The greatest advantage of the LBM is its high efficiency for parallel computing [73]. Hence, the LBM is considered to be more efficient, and highly recommended for the simulation design of complex geometric structures. In addition, CFD methods have been known to be macro-scale simulations, and thus, they have limitations for micro-scale calculations, such as small-size particles at the nano or micrometer scales, including bacteria, viruses, or fungi suspended in air. These small biological particles can become suspended in air by gravitational, Brownian and interaction forces, and they also have a penetration rate under air flow. Recently, the LBM has been shown to be a meso-scale simulation method with good parallelism, robustness, clear phase boundaries, and fast convergence for theoretical and computational approaches [73]. However, the recent application of the LBM in fluid simulations only involved one or two-dimensional analysis, including indoor air or solid/liquid binary-phase systems [39,74-76]. Although CFD models have been widely used recently in the dynamic simulations of personal-inhalable bioaerosols and for risk analysis of infectious airborne diseases, improved three-dimensional-phase, accurate, and fast-caulculated LBM models with clear boundaries will be imperative to simulate the integrated parameters in triple-phase systems, including air, particles and liquid media, and to guide the optimization of bioaerosol cyclone design with a highest sampling efficient selection in the future study.

Determining design parameters for high bio-efficiency of cyclones

To reach the ultimate goal of an ideal cyclone, the bio-efficiency is always the most important aspect of bioaerosol sampling. The three formulas for bio-efficiency (η_AA, η_AH, and DE) described above are appropriate for comparing a family of geometrically similar cyclones and can be applied for the modification of specific units in cyclone re-design under the same restricted conditions. In addition, convergent variables that are more comprehensive and wider scope associated with these parameters are needed to provide performance and design evaluations.

In recent studies, the design of the aerosol and liquid sampling inlet device with a proper air flow rate is essential for airborne particle collection. Haig et al. [28] suggested that the inlet of the cyclone is better positioned horizontally to prevent sampling bias. Once the air sample is taken-in by the device, the path of the airborne particle flow should be as straight and direct as possible to prevent losses in the bends of the conducting tube. Furthermore, the width of the inlet entrance was found to be a critical parameter for a WWC [16]. McFarland et al. [32] re-designed a WWC with four modifications that enabled a particle AD range from 1 µm to 10 µm collection. The modifications of the upgraded device were the flow path, inlet slot, cyclone body length and liquid flow rate [10].

Ideally, for high aerosol collecting efficiency, sampling inlet, transport and deposition of a cyclone are three main aspects that should be focused on when designing a sampling system. Each can be assessed based on the specific components (Table 2). Correlations evaluated from numerical predictions (three-dimensional-phase LBM module), experimental measurements (inactive bacteria aerosol simulated house), and field studies (natural ventilation) provide results with a high degree of confidence to aid in the optimization of the design parameters (Figure 7). Integrating the data from the numerical simulations, wet-lab calibrations, and microbiological aerosol demonstrations would provide a comprehensive analysis for the many complex designing parameters to obtain an ideal, high-sampling, efficient cyclone.

Figure 7. Integrated scheme for development of an ideal design of a cyclone.

Table 2. Listing of various geometric design units for the cyclone sampling system.

	Design parameters of sampling system components
Main parts	Sampling inlet	Sampling transport	Sampling deposition
Detailed units	Air flow rate Inlet shape: orientation, width, angles Particle size Pressure drop	Tube length Bend angle Transit time Flow type (laminar/turbulent) Materials of tube and probe Contraction on tube	Liquid flow rate Film thickness Shape of collecting container Liquid medium

Future design and trend of cyclones

Need for real-time bioaerosol sampling and detection

With advances in technology, real-time bioaerosol sampling and online detection have become increasingly important for a wide variety of research communities and scientific questions [45]. There is a need for improved, affordable, real-time, high-sampling-efficiency detection instruments that measure particle types at sufficiently high resolution with an accurate specificity. The combination of high-efficiency cyclone sampling and real-time bioaerosol sensing techniques, including fluorescence spectroscopy, microfluidic techniques or quantitative polymerase chain reaction (qPCR), will provide a fast response with adequate counting and discriminative results.

Need for improved links between numerical modelling and experiments

Cyclone behavior is complex and difficult to model. In recent years, many researchers have developed high-sampling-efficiency predictive models for cyclone performance, which provide insight into how cyclone performance is affected by the geometric design [30]. Detailed experiments of new numerical modelling prediction insights will provide confirmed data for improvements such as an LBM model. Continued and increased communication between the modeling and experiments is encouraged so that cyclone development and monitoring efforts are focused on the collection of solid datasets that can most efficiently reduce numerical model uncertainties.

Need for database development

Open databases of the performance parameters for different types of cyclones should be developed for the techniques listed in Table 2. These should include as many bioparticle sizes as possible, aerosolization parameters and clearly reported analysis conditions. This database should open a wide range of scientific objective and cross communities, including a cyclone’s geometric design based on high-collecting efficiency optimization by data mining or artificial intelligence. Openly shared results and big-data platforms for cross-disciplinary interactions will provide more opportunities for cyclone research.

Conclusions

The development of well-designed cyclones with high sampling efficiencies, fast and continuous sampling functions, a wide range of particle cut-off diameters, and good microorganism viability has progressed considerably in recent years, with a broad range of techniques being established or under development. As indicated by the overview provided in this review, the aerosol-to-hydrosol efficiency is intrinsically related to the cyclone performance parameters that are useful for design. However, the challenges for the optimization of cyclone performance parameters are still far from being integrated for real-time sampling and detection to link numerical modelling and experiments, and to develop a database. These challenges will be advantageous for the further development and refinement of cyclone sampling efficiency research and geometric design, and they will continue to open new opportunities for routine bioaerosol monitoring and research.

Disclosure statement

No potential conflict of interest is reported by the authors.

Additional information

Funding

This work was supported by the Chinese Academy of Inspection and Quarantine under Grant No. 2018JK015.