(465i) Exploring Aerosol-Gas Dynamics in Fluidic Logic Oscillator Ventilation

Infectious diseases such as COVID-19 and wartime threats of biological or chemical attacks cause respiratory depression or failure, diminishing the effectiveness of military units. Safe and effective mechanical ventilation is the single most important skill performed by emergency responders but is not easily mastered due to a myriad of reasons. Bag valve masks (BVM) require appropriate ventilation pressures and tidal volumes, as well as the ability to titrate oxygen. Mechanical ventilators, while more efficient than BVM, require electricity. They are better at minute ventilation, safe pressures/volumes, and gas exchange. They enable providers to accomplish tasks, document more completely, and provide better care.

The Fluidic Logic Oscillating Ventilator (FLOV) is the world’s smallest and spontaneously breathing ventilator. It is gas driven, user friendly, easily manufacturable, automatic, and it provides repeatable and clinically acceptable parameters with only part-time clinician involvement. The FLOV relies on a source of motive oxygen air of ≤ 20 L/min. Boundary layer separation, impingement on a “splitter,” and jet deflection cause the flow to cyclically “flip” between inhalation and exhalation without the need for moving parts or any other intervention. For the first time, we investigate the possibility and ramifications of aerosol inhalation therapy utilizing the FLOV. Aerosols can be used to treat/administer bronchodilators, corticosteroids, antibiotics, pulmonary surfactant, mucolytics, biologicals, genes, proteinoids, transpulmonary cooling and other life-saving procedures.

Background

The original work this paper is founded upon is:

“The Nature of ‘Searching’ Vortices in Fluidic Logic Driven by a Switching Jet,” J. Fluids Eng., vol. 144, no. 8, Mar. 2022, doi: 10.1115/1.4053786.

This paper, written by Wayne Strasser Ph.D., P.E., focuses on the gas dynamics within the FLOV. The study involved a 2.6 million element mesh with 99.97% (by volume) structured hexahedral cells of the FLOV geometry. The typical mesh element length scale is 30 microns. In the study’s simulations, compressible gas phase governing relations in vector notation are solved using commercial code ANSYS Fluent. These equations are listed below:

∂ρ/∂t + ∇ * (ρu) = 0

∂/∂t(ρu) + ρu * ∇u = ∇ * (τ + τ_f) - ∇P + ∑F

∂/∂t(ρC_pT) + ∇ * [u(ρC_pT)] = ∇ * (ζ∇T)

It should be noted that each ∇, u, τ, τ_f, and F should be evaluated using their respective average values. All properties describe the gas phase only, so phase subscripts are removed. t is time, u is the velocity vector, τ is the laminar shear stress tensor, τ_f is the implicitly filtered (ILES) shear stress tensor, P is static pressure, ∑F is the summed droplet-gas force vector (important discussion in the next section), C_p is constant pressure heat capacity, T is the absolute static temperature, ζ and is laminar thermal conductivity. An attached lung is simulated via user-defined functions (UDF). This lung does not impose a certain flow cycle; it only acts as a balloon. The ventilator is designed to even work on an unconscious patient, so the flow direction changes are spontaneous in the real ventilator and CFD. The boundary condition static pressure at the patient port of the FLOV is therefore defined by:

P(t) = φ(t)R + ∀(t)/C

Where R is the lung and connector piping flow resistance factor in units of cmH2O/(L/s), φ(t) is the instantaneous volumetric flow out of or into the lung port, ∀(t) is the instantaneous lung volume, and C is the lung compliance in units of ml/cmH2O. The paper compares its CFD results with those of a 3D printed FLOV; the CFD results correlate very closely with those of the printed FLOV; therefore, the flow pattern of the model is confirmed, and research can be advanced into the investigation of the effects flow has on aerosol particles, and vice versa.

Methods

One of the main concerns while studying the feasibility of aerosol inhalation therapy through the FLOV are the effects of the flow on droplets (“feed-forward”). Uniform distribution in the patient delivery space as well as time/evaporation of the particles is desired. Another concern is the effects of droplets on the flow (“feedback”). We investigate feed-forward effects herein. The following equation represents feed-forward physics, describing gas flow influences on droplet motion and potentially creating preferential concentration.

m_jdu_d,j /dt = F_d,j F_L,j

This equation only focuses on drag and lift forces; droplet-droplet collisions are ignored due to a low droplet concentration.

A continuous injection of inert particles is initiated into the verified FLOV geometry. Between several simulations, these droplets have uniform 2.5, 5, and 10 µm diameters with a 1∗10^-20 kg/s flow rate (too low for feedback and collisional effects). Droplets have a density of 1000 kg/m³. To understand the results of these simulations, there are three important reference regions to consider. As seen in Figure 1, the droplet inlet is leftmost region the particle tracks originate from. The outlet leading to the patient is the upper part of the geometry on the right where the particles continue to flow. The outlet to atmosphere is the bottom region of geometry with no droplets. Concentration is measured at both outlets.

Results

Results are gathered for all droplet diameters to assess the distribution of droplets when affected by the flow, shown by normalized DPM concentration (NDPMC). DPMC is normalized due to such low concentrations and is done by the dividing the computer concentrations by the average 3.33∗10^-17 kg/m³. Any concentration above the average (1.0 on normalized plots) indicates non-uniform distribution. In Figure 2, the maximum NDPMC value is indicated for each droplet diameter every 5 time steps over two flow cycles. From the plot ordinate, it is obvious that concentrations locally across the two outlets exceed the mean by up to two orders of magnitude. Droplet diameter has a clear effect on these results; the average maximum NDPMC values are approximately 12.7, 21.2, and 15.1 for the 2.5, 5, and 10 µm diameters, respectively. This trend can be observed in Figure 2. Additionally, particle size influences the time droplets have their largest impact. In the same plot, a small shift along the x-axis can be seen between the data sets that grows larger by the second flow cycle. The 10 µm droplets complete the inhalation transition approximately 0.5 cycles sooner than the 2.5 µm particles, with the 5 µm data falling in between. Particle tracks colored by residence time are also developed. Droplets are tracked for a maximum of 0.00275 seconds or 1250 steps, with white paths indicating the largest time. If large amounts of particles do not completely pass through the unit within that duration, they are likely hung up in the geometry (vortex trapping). Figure 1 displays the droplet path as the patient inhales. Very similar results (with slightly varying residence times) were found for all droplet diameters.

Implications

It is concluded that droplets are mal-distributed to the patient, and the highest residence times for particles are found during transitions in the ventilator’s natural breathing. In the future, parameters for feed-forward investigation will continue to be iterated. Droplet diameter and mass flow rate will be altered, along with numerical sensitivity analyses, to more clearly analyze the flow’s effect on the droplets. For feedback effects, the aim to produce a droplet size/concentration “safety” map to help users define a safe operating range for administering treatment via the FLOV.