Real Breathing Waveforms Reshape How Aerosols Deposit in Patient-Specific Airways

Every breath a person takes is a carefully choreographed asymmetry: inhalation is shorter and sharper, exhalation is longer and slower, and the air rushing through the nose, throat, and bronchial tree never behaves quite the way simplified models assume. A new computational study published in Results in Engineering has now mapped, second by second, how that asymmetry changes where inhaled particles actually land inside a realistic human airway. The work, led by Subho Samanta and Suman Chakraborty at the Indian Institute of Technology Kharagpur together with colleagues, couples a CT-reconstructed airway geometry with a spirometry-derived breathing waveform and tracks tens of thousands of individual aerosol particles through a complete breathing cycle, revealing deposition dynamics that inhalation-only and steady-flow simulations have long missed.

The airway model at the heart of the study is an anatomically realistic reconstruction originally derived from computed tomography scans of an 84-year-old woman, spanning the external facial region through the nasal cavity, pharynx, larynx, trachea, and roughly nine generations of the bronchial tree, with 135 terminal outlets representing the distal bronchi. Rather than injecting particles as a single batch at the start of inhalation, as many earlier studies have done, the researchers released 30,000 particles continuously at every simulation timestep throughout the inspiratory phase, synchronizing particle entry with the transient breathing waveform itself. Three particle diameters were chosen to span clinically and environmentally important regimes: 1 micrometer for fine pharmaceutical aerosols and ambient fine particulate matter, 5 micrometers for the therapeutic window of metered-dose and dry powder inhalers, and 10 micrometers for the coarse, inertia-dominated fraction filtered heavily by the upper airways.

The computational machinery behind the simulations is considerable. The team used a Cartesian cut-cell meshing approach in the CONVERGE CFD solver, refining the grid near airway walls down to roughly 0.11 millimeters within the bronchial tree, for a final mesh of about 2.7 million control volumes. A formal grid convergence study using the Grid Convergence Index methodology confirmed that the solution was essentially mesh-independent, with the convergence index falling to just over 1 percent at the adopted resolution. Airflow was treated as incompressible and predominantly laminar, an assumption the authors justify by the modest mean inspiratory flow rate of 12.7 liters per minute, noting that transitional conditions arise only at the instantaneous peaks of the breathing waveform, where laryngeal Reynolds numbers reach about 3,740 during inhalation and 6,268 during exhalation.

Particle motion was resolved with a Lagrangian tracking framework in which each particle’s trajectory is integrated individually through the unsteady airflow field. The force balance included aerodynamic drag evaluated with Clift-Grace-Weber correlations, gravitational settling, and shear-induced lift modeled with the Saffman-Mei correlation. Brownian motion was neglected because the particles considered are all one micrometer or larger, and inter-particle collisions were excluded given the dilute nature of inhaled aerosols. Airway walls were assigned trap conditions, mimicking mucus-lined surfaces where contacting particles deposit permanently, while particles exiting the distal outlets during inhalation were classified as having escaped into deeper lung regions beyond the truncation plane. A rigorous particle-number balance confirmed that deposited, escaped, exhaled, and suspended particles summed exactly to the 30,000 injected in every case.

Validation proceeded along three independent lines. Predicted velocity profiles in a nasal cavity matched the classic hot-wire anemometry measurements of Hahn and colleagues under restful breathing conditions, and a benchmark study by Li and co-workers had already established that laminar modeling outperforms all turbulence closures at that flow rate. Predicted pressure drops from the airway inlet to the pharynx agreed closely with independent numerical results across steady flow rates of 5 to 15 liters per minute, and deviations from experiments on 3D-printed nasal replicas were consistent with known printing roughness and measurement uncertainty. Finally, extrathoracic deposition fractions fell within the scatter band of six independent in vivo and numerical datasets across nearly four decades of the impaction parameter, confirming that the framework captures the dominant inertial deposition physics.

The headline finding concerns the breathing waveform itself. When the realistic spirometry-derived profile, with its asymmetric timing and inspired volume of 337.52 milliliters, was compared against an idealized sinusoidal waveform of the same mean flow rate but a larger inspired volume of 423.02 milliliters, the sinusoidal case produced one-cycle deposition fractions of 53.1, 66.1, and 69.7 percent for 1, 5, and 10 micrometer particles, against 45.9, 56.0, and 61.4 percent under realistic breathing. That corresponds to relative differences of roughly 14 to 18 percent, with the largest gap at 5 micrometers, the size where inertial impaction and gravitational settling contribute comparably. Because the two waveforms differ in both temporal shape and inspired volume, the authors caution that the individual contributions of each factor cannot be separated in this comparison, but the message is clear: the assumed breathing profile alone can shift deposition estimates by double-digit percentages.

Perhaps the most striking result is what happens during exhalation. Under the spirometry-derived waveform, between 12.9 and 15.0 percent of all particles deposited within the cycle did so during the expiratory phase, as flow reversal, secondary motions, and renewed impaction at the pharyngeal and laryngeal constrictions drove suspended particles onto airway walls. Under sinusoidal breathing the exhalation contribution was smaller, at 8.0 to 9.1 percent, largely because fewer particles remained airborne at flow reversal. The instantaneous deposition rate showed distinct secondary peaks during early expiratory acceleration, reaching up to nearly 4,935 particles per second for 10 micrometer particles. Any simulation restricted to the inhalation phase would be blind to this entire deposition mechanism, which represents drug lost to proximal re-impaction for therapeutic aerosols and previously uncounted dose for toxic ones.

The study also quantified how the airway wall experiences breathing. Borrowing metrics from cardiovascular fluid dynamics, the team computed the time-averaged wall shear stress and the oscillatory shear index across the airway surface. Shear loading was consistently highest in the pharynx and larynx, where geometric constrictions accelerate flow, and was generally stronger and more spatially extensive during inhalation than exhalation. The oscillatory shear index, which flags regions of frequent flow-direction reversal, rose during expiration in the supraglottic and pharyngeal regions, precisely where the exhalation-phase deposition peaks occurred. Because high oscillatory shear implies prolonged near-wall particle residence, these maps provide a mechanistic bridge between unsteady flow structure and the spatial pattern of deposited dose.

Comparisons against the widely used Multiple-Path Particle Dosimetry model exposed the limits of population-averaged approaches. For 10 micrometer particles, both the CFD simulation and the MPPD model predicted near-complete deposition, at 97.7 and 95.9 percent respectively. But at 5 micrometers the CFD total of 69.9 percent fell well below the MPPD prediction of 98.6 percent, largely because particles escaping the truncated computational domain would deposit in smaller airways beyond it, while at 1 micrometer the CFD predicted 55.7 percent against MPPD’s 35.3 percent, reflecting how this specific subject’s morphology retains more fine particles than an average-adult model predicts. The deviations run in opposite directions at different sizes, underscoring that individual anatomy and domain coverage matter in ways one-dimensional models cannot capture.

The implications reach across environmental exposure science, inhalation toxicology, and respiratory medicine. For occupational and ambient exposure assessment, the results show that waveform choice alone can alter predicted deposition of dusts, spores, and fine particulates by 13 to 18 percent. For drug delivery, the 10-percentage-point difference in 5 micrometer deposition between waveforms bears directly on the design and evaluation of metered-dose and dry powder inhalers, particularly next-generation carrier systems engineered for site-specific deep lung delivery, where the fraction of aerosol surviving proximal filtration determines therapeutic efficacy. The authors are careful to note the study’s scope: it uses a single elderly female geometry, rigid walls, uniform pressure outlets, and laminar assumptions, and the two waveforms were matched in mean flow but not inspired volume. Extending the framework to anatomically diverse cohorts, incorporating airway compliance and impedance-based outlet conditions, and quantifying turbulence sensitivity at expiratory peaks are identified as priorities. Even so, the demonstration that realistic breathing dynamics, subject-specific anatomy, and full-cycle particle tracking together reshape deposition predictions marks a concrete step toward personalized respiratory dosimetry, in which an individual’s own airway anatomy and measured breathing pattern, rather than population averages, determine where each inhaled particle ultimately comes to rest.

Subject of Research: Time-resolved computational fluid dynamics of airflow and aerosol deposition in CT-reconstructed human airways under realistic breathing waveforms

Article Title: Patient-specific time-resolved mapping of airway flow and aerosol deposition using CT-derived geometries

Article References: Samanta, S., Dwivedi, S., Nair, A. G., Kolliyil, J., Hirani, H., & Chakraborty, S. (2026). Patient-specific time-resolved mapping of airway flow and aerosol deposition using CT-derived geometries. Results in Engineering, 32, Article 113294. https://doi.org/10.1016/j.rineng.2026.113294

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113294

Keywords: aerosol deposition, computational fluid dynamics, CT airway reconstruction, breathing waveform, Lagrangian particle tracking, inhalation therapy, respiratory airways, wall shear stress, drug delivery, exposure assessment, spirometry, lung dosimetry