The air flowfield generated in the Aerolizer® for the different cases studied was simulated using the commercial CFD code ANSYS CFX5 (versions 5.6, 5.7 and 5.7.1). The angular velocity of the capsule, required to model the motion of the rotating capsule, was determined using high-speed photography. The computational results were validated using Laser Doppler Velocimetry (LDV) techniques. Experimentally, the dispersion performance of the inhaler was determined using a spray-dried mannitol powder (particle size d50 = 3.2um, span [(d90-d10)/d50] = 1.3) and a 4-stage liquid impinger. The dispersed mannitol was assayed by high performance liquid chromatography, using refractive index detection, which allowed the mass of powder deposited at each location and fine particle fraction to be determined. In this thesis, the fine particle fraction was defined as the mass fraction of particles smaller than 5µm, referenced against either the total mass of powder loaded into (FPFLoaded), or the total mass of powder emitted from (FPFEm), the device.
The results showed that the design of a dry powder inhaler can significantly affect the flowfield generated in the device and subsequently, the overall inhaler performance. The structure of the grid was found to significantly affect the performance of the device. As the grid structure was reduced, the amount of powder retained in the inhaler increased from 17% to 34% (due to increased tangential flow of particles in the inhaler mouthpiece) and the FPFLoaded reduced from 47% to 35% (due to increased mouthpiece retention). The length of the mouthpiece played a lesser role on the inhaler performance, having no effect on the flowfield generated in the device or the amount of throat deposition. Only the amount of device retention varied as the mouthpiece length was reduced.
The device air flow rate had a significant effect on the inhaler dispersion performance by controlling the deagglomeration potential of the device flowfield. Both powder dispersion and throat deposition increased with air flow. The amount of capsule retention decreased with flow, while device retention first decreased then increased with flow. The optimal inhaler performance was found at 65 l min-1 showing a high FPF of 63% with low throat deposition (9.0%) and capsule retention (4.3%). CFD analysis of the device flowfield at this critical flow rate has allowed us to determine the distribution of integral scale strain rates (an important characteristic of the turbulent flowfield responsible for particle break-up) and average particle impaction velocities that maximised the inhaler dispersion performance.
The air inlet size had a varying effect on powder dispersion at different flow rates. At low flow rates (30 and 45 l min-1), reducing the air inlet size increased the inhaler dispersion performance by increasing the flow turbulence levels and particle impaction velocities above their critical levels for optimal powder dispersion. At 60 l min-1, reducing the air inlet size reduced the inhaler dispersion performance by releasing a large amount of powder from the device before the turbulence levels and particle impaction velocities could be fully-developed. The results demonstrated that the optimal inhaler dispersion performance can be predicted if details of the device flowfield are known and highlighted the importance of minimising the amount of powder released from the device prior to full flow development.
Finally, the geometry of the inhaler mouthpiece had no effect on device retention or the inhaler dispersion performance. In contrast, the mouthpiece geometry strongly affected the amount of throat deposition by controlling both the exit air flow velocity and the radial motion of the emitted aerosol. Upon varying the inhaler mouthpiece geometry from a cylindrical to conical design, a balance between reduced throat deposition (due to reduced exit velocity) and increased throat deposition (due to increased radial aerosol motion) was found. Despite a reduced throat deposition, there was no difference in the overall performance of the inhaler, suggesting that the Aerolizer® performance may not be significantly improved by modifying the design of the inhaler mouthpiece.
The work performed in this thesis demonstrates the effectiveness of using computational fluid dynamics to simulate the flowfield generated in dry powder inhalers of different design. The computational results were used to provide fundamental information to explain differences in the observed dispersion behaviour. The combination of computational and experimental methods was used successfully to study the factors affecting agglomerate dispersion in dry powder inhalers. In summary, the performance of a dry powder inhaler can be affected by a number of inter-relating factors. CFD combined with experimental results provides a rational basis for understanding the performance differences. The approach adopted here is readily applicable to other dry powder inhaler systems to help better understand their performance optimisation.