Effect of an upstream grid on the fluidization of pharmaceutical carrier powders

Abstract

The influence of grid generated mixing on the fluidization of pharmaceutical carrier powders is studied in a channel-flow experiment using direct high-speed imaging and particle image velocimetry (PIV). Four different lactose powders with mass median diameters that range between 61μm to 121μm are used. The degree of powder mixing in the flow as a function of grid position relative to the powder bed and grid area blockage ratios (ranging from ~25% to ~40%) is studied for a range of flow-rates. The study presents comprehensive mappings of how pharmaceutical powders are fluidised under the influence of mixing, by examining powder bed morphology, powder emptying rate, and the local flow-field surrounding the pocket. The use of a grid results in higher evacuation percentages (void fraction) and a faster evacuation rate but is associated with randomized evacuation behaviour as observed from the powder bed morphology. Use of a grid can enable evacuation of powder at lower overall flow-rates, which may have important implications on respiratory drug delivery. PIV results show the trend of mean velocities with the mass median powder diameter and demonstrates how a grid with lower blockage ratio can increase the degree of mixing of the evacuating powder and make the evacuation process more rapid. This study contributes towards a better understanding of fluidization processes as relevant to dry powder inhaler devices and sheds light on how simple design alterations, such as adding an upstream grid, can be incorporated to optimise device effectiveness.

1. Introduction

Fluidization of pharmaceutical powder from dry powder inhalers (DPIs) is a critical step towards producing an inhalable mixture for delivery of small particles to the lungs. Carrier particles are a central part of this, as they provide the necessary lift force to enhance fluidization (Islam and Cleary, 2012), and ensure that the active pharmaceutical ingredient (API) attached to the carrier is able to evacuate from the inhaler device effectively. A significant volume of work has focused on designing powder formulations with improved fluidization and dispersion characteristics (Islam and Cleary, 2012), focusing for instance on improving the formulation of the carrier particle by adding fine particles or a ternary age (de Boer et al., 2012; Guenette et al., 2009; Iida et al., 2000; Malcolmson and Embleton, 1998), altering the carrier size (Guenette et al., 2009; Kaialy et al., 2012), or altering the morphology of the carrier particle (Kaialy et al., 2011). Whilst formulation modifications are useful to improve the function of dry powder inhalers, understanding critical mechanisms to fluidise the power within the device effectively as a function of flow is equally important(Cheng et al., 2019; Coates et al., 2005; Islam and Cleary, 2012; Longest et al., 2013).

Experimental work in this space has been performed by researchers to understand powder evacuation mechanisms and the role of turbulence. The study by Voss and Finlay (Voss and Finlay, 2002) showed the importance of turbulence and mechanical impaction on dry powder de-agglomeration using an impaction rig and Diskhaler in which laser Doppler velocimetry was used to measure flow turbulence. The study by Voss and Finlay (Voss and Finlay, 2002) suggested turbulence as an effective de-agglomeration mechanism, though full isolation and control over the turbulence intensity was not possible. Work by Tuley et al. (Tuley et al., 2008) investigated the fluidization of four different powders in three simplified geometries to assess the effect of powder type, geometry, and inhalation profile on powder fluidization using high-speed imaging. Fracture and erosion mechanisms were identified as the two basic modes for evacuation, with their contribution depending on powder type. In a similar type of study, Versteeg et al. (Versteeg and Wildman, 2004) studied the fluidization of powder concluding that jet impaction was a dominant mechanism for powder evacuation and suggested that being able to control jet impaction is useful to improve DPI design. Recent work by Mahmoudi et al. (Mahmoudi et al., 2019) focused on using a two laser beam extinction method to characterize lactose powder evacuation in a simple channel flow in which the powder pocket was subjected to a shear flow at turbulent Reynolds numbers. At lower Reynolds numbers, the evacuation process was sensitive to the powder properties, however higher flow rates saw an increasingly irrelevant role of properties on the total evacuation time of the powder.

In addition to studying basic powder transport mechanisms in either inhaler devices (Coates et al., 2006; Coates et al., 2004) or simpler geometries (Tuley et al., 2008), research has also focused on understanding the role of specific design features of inhaler devices. One such key feature is a grid, which is commonly found just upstream of the outlet of a DPI mouthpiece and exists in order to enhance de-agglomeration though mechanical impaction. Coates et al. (Coates et al., 2004) investigated the effect of grids on the performance of an Aerolizer using laser Doppler velocimetry and computational fluid dynamics (CFD) with a key conclusion that the grid structure affects DPI performance, powder retention time and fine particle fraction. Wong et al. (Wong et al., 2011) demonstrated the effect of grid structure on the aerosolisation and agglomerate break up through a standardised entrainment tube. The flow was simulated using CFD while laser diffraction and chemical analysis were utilised to assess deposition. Longest et al. (Longest and Farkas, 2019; Longest et al., 2019) also recently conducted aerosolization experiments and CFD to better predict the performance of a newly proposed DPI. The effect of inlet and outlet orifices was studied with results suggesting that increasing turbulence leads to an increase in the emitted dose (ED). Interestingly however, low turbulence caused a favourable decrease in the mass median aerodynamic diameter (MMAD).

It is evident that research to date has primarily focused on the characterisation of specific inhaler devices, with the motivation to define device performance. Nevertheless, description of the fundamental factors that cause powder deagglomeration and dispersion, in particular the independent effects of turbulent mixing, is lacking in the literature (Kourmatzis et al., 2018). A turbulence generating grid placed upstream of a powder pocket can be used to control local mixing and increase the variation in local velocity. This is the approach used in this contribution to fill the knowledge gap (Chan, 2006) of how increased fluctuations in the gas phase can be advantageously used to control powder evacuation behaviour. This study isolates the role of grid generated fluctuations on the fluidization behaviour of pharmaceutical lactose powders over a range of turbulent Reynolds numbers, as relevant to DPI flows(Islam and Cleary, 2012). Carrier powders with a range of particle properties are used, resulting in a range of cohesion and particle sizes. A simple fully-developed channel flow is used in which boundary conditions are kept simple, and grids utilised over a range of flow rates to alter the powder fluidization process. Fundamental information from this type of experiment can be used to understand the role of an upstream grid in pharmaceutical powder fluidization whilst also contributing to the development of dry powder fluidization models. Direct high-speed imaging, followed by image processing, and particle image velocimetry (PIV) is used to measure and analyse the evacuation time, evacuation percentages (void fraction), and particle velocities (mean and fluctuating) emerging out of the powder pocket as a function of flow conditions and powder properties.

2. Experimental methodology

The fluidization channel is shown in Fig. 1. The region of interest here is the powder pocket that has a rectangular shape with width (W) and pocket depth (D). The channel cross-sectional area is square, with the side length being equal to the pocket width (W). Two channels with different pocket widths of 5 mm (5×5 channel) and 10 mm (10×10 channel) were used. The channel depth is kept constant (D=5mm) in both channels. The powder pocket is located at a distance (L₁=250 mm) downstream of the inlet allowing for a developed flow with air supplied to the channel through a control valve, pressure gauge, and rotameter, allowing for control of the flow rate of up to 100 litres/min. A solenoid valve (12V DC) is used to control the start of the air-flow through the system, which remains fully open until the powder evacuation process is completed. This experimental configuration follows the same methodology described elsewhere (Mahmoudi et al., 2019).

Figure 1.

A) Schematic of experimental setup showing layout for PIV and high-speed imaging. Laser sheet (used for PIV only) is positioned at the middle of the channel width and across the powder pocket. A grid (see Panel B) is placed upstream (see Panel C) of the powder bed. B) Dimensions and geometry of a grid with blockage ratio of ~30%. C) This panel shows the side view of the horizontal channel and the dimensions of the channel, powder bed and location of grid relative to the powder bed.

Figure 1 shows an example of one of the grids with a mesh size (M) of 1.92 mm. Two other grids of mesh sizes 1.6 mm and 2.4 mm were also used in the 10×10 channel. The blockage ratio of the grids is approximately ~25%, ~30%, and ~40%, and is calculated as the area occupied by the grid normalised by the channel cross-sectional area. These values were chosen to span a range of blockage ratios within the physical constraints of manufacturing the grids. Due to physical limitations to implement the grid in the 5 × 5 channel, grids were only tested in the 10×10 channel. The 3D printed grid is placed before the powder pocket at a distance (X), as shown in Fig. 1. The degree of mixing in the vicinity of the pocket is studied as a function of X (0, 2, 5, 8, 10, 15, and 20 mm).

2.1. Imaging Layout

Two experimental techniques are employed in this study, and they are 1) direct high-speed imaging of the powder bed to obtain information on powder evacuation rates and powder bed morphologies and 2) particle image velocimetry (PIV) for velocity measurement. For both high-speed imaging and PIV, images were acquired using a Photron FASTCAM Mini UX100 high-speed camera with 8GB internal memory. For high-speed imaging, the images were acquired at frame rates that ranged from 250 fps to 20000 fps depending on the flow rate and channel size (10 ×10 or 5 × 5). The camera shutter speed was set to 1/100000 sec for all the cases. For the 5 × 5 channel, 92 pixels square were used to image the pocket, and for the 10 ×10 channel, 184 × 92 pixels were used, resulting in a spatial resolution of ~54.34 μm. The front lens was a 28 mm OZUNON camera lens with 52 mm diameter. The images were stored using Photron’s FastCAM viewer 3 software. The video recording is initiated using a background subtraction trigger applied to the powder bed surface in which the background images are created by taking average brightness value per pixel for 100 images before evacuation starts. The live images are compared to the background images, on a pixel by pixel basis, and image acquisition begins if the change exceeds a specific threshold. This ensures that image acquisition begins precisely at the start of the powder evacuation process.

For the high-speed imaging studies, a continuous light source of 500W was used in a side illumination arrangement to illuminate the front of the powder pocket. An in-house image processing script was utilized for image processing which converted the captured images into a binarized form in order to identify the powder-air interface. An adaptive binarization threshold was utilized such that the top threshold used was a percentage of the maximum intensity of the image. The threshold value used varied between 40% to 55%, and this is in excellent agreement with recommended background thresholds for binarizing images from similar two-phase flows for droplet size measurement, which has been previously validated against phase Doppler anemometry data (Kourmatzis et al., 2017).

For PIV measurement, an Oxford Lasers Firefly pulsed diode laser with an average power of 300W was used to produce the laser sheet that passes through the middle plane of the channel, as shown in Fig. 1. In-built sheet forming optics were utilized to generate a sheet over the powder bed. The laser diode is controlled by a laser control unit and is synchronized with the camera by connecting the camera output to the laser control unit together with the solenoid valve, which controls the flow and is actuated by the trigger signal of the high-speed camera. PIV cross-correlation post-processing code (PIVLAB (Thielicke and Stamhuis, 2014)) was utilized to generate particle velocity vectors. The PIV performed here uses a fast Fourier transform (FFT) window deformation multi-pass algorithm using an interrogation window with a final size of 16 × 16 pixels with 50% window overlap for computing the cross-correlation. Interrogation deformation is calculated using the displacement information with bilinear interpolation. A threshold for acceptable velocities was applied to remove outliers from the vector field via graphical selection. Two times 3-point Gaussian fit is used for refining the location of the intensity peak of the correlation matrix.

2.2. Powder properties and powder loading

Four lactose carrier powders (DFE Pharma) were used in the current experiment. Two from the Lactohale® series (LH200 and LH206) which are milled lactose powder and two from the Respitose® series (SV003 and SV010) which are sieved lactose powder. The properties of both these powders are shown in Table 1. The milled powder has a larger particle size span (×90 − ×10)/(×50) compared to the sieved powder. LH200 has the highest percentage of fine particles compared to the other powders and the highest Carr’s index of 39 inferring poorer flow-ability and higher cohesion. The Carr’s index, provided by the supplier (DFE Pharma) is determined from the poured and tapped bulk density of the powders, measured according to method 616 of the United States Pharmacopeia. SV010 has the largest mass median diameter (112 μm), while SV003 has the lowest mass median diameter (61.6μm). Readers interested in the details of the particle size distributions, and other properties of these powders are directed to the authors’ previous work (Mahmoudi et al., 2019). Extreme care was taken to ensure that powder loaded into the powder bed was not packed, and the excess powder was swiped away. The loading method is similar to that used elsewhere (Mahmoudi et al., 2019; Tuley et al., 2008) and has been proven to be very repeatable, with an analysis of the error associated with this type of repetitive measurement being presented in the Appendix of (Mahmoudi et al., 2019).

Table 1.

Key properties of powders used in the experiments where ×10, ×50 and ×90 are the diameter intercepts for 10%, 50% and 90% of the cumulative volume determined from the cumulative particle size distribution, %<Nμm indicates the percentage of fine particles below N μm

	Lactohale® 200	Lactohale® 206	Respitose® SV003	Respitose® SV010
Tapped density (1250 taps) (g/l)	950	720	780	830
Poured density (g/l)	580	870	630	690
Carr’s index %	39	17	19	17
x10 (μm)	10.7	32.0	32.3	51.0
x50 (μm)	74.5	84.9	61.6	112.0
x90 (μm)	158.6	166.2	95.7	181.1
Span=(x90-x10)/x50	1.98	1.58	1.03	1.12
%<5 μm	4.8	2.3	2.8	1.6
%<10 μm	16.2	4.1	5.4	3.8
%<15 μm	28.2	6.6	8.6	5.8

3. Results and Discussion

The first part of the results focuses on high-speed images of the powder pocket with corresponding discussions on powder bed morphologies and evacuation rates for different turbulence generating grids and their positioning with respect to the powder pocket. The second part of the results presents PIV measurements of the area above the powder pocket for the 10 ×10 channel. Mean and fluctuating values of velocity are also presented and evaluated to isolate the effect of an upstream grid on particle dynamics in the powder pocket region.

3.1. Qualitative powder bed morphology

The morphology of the powder bed for the 5 × 5 channel for a case where no grid was used, as a benchmark case, is shown in Fig. 2(a). Images for the powder pocket at different time instances are binarized, and the powder-air interface is subsequently identified as discussed in section 2.1. An example of an identified interface for a 10 ×10 channel case with no turbulence generating grid is shown in Fig. 2(b). As shown in Fig. 2 (a–b), without using a grid, evacuation starts at the upper-left edge of the powder pocket with the airflow over the powder pocket in this case being from right to left. This type of powder emptying behaviour has been observed previously in (Tuley et al., 2008). Evacuation commences with a small recirculation zone generated at the upper left edge that grows with time. These evacuation patterns are consistent for all other powders used in this study and correspond well with previous work (Mahmoudi et al., 2019) which also shows that the behaviour is independent of the channel geometry size. Using a grid completely changes the powder bed morphology from a regularly defined recirculation zone to an evacuation process that is random such that it does not necessarily start at the top-left edge (see Fig. 2(c) for LH206 and SV010 at 80 L/min, using a grid with a blockage ratio of ~30% with X = 0 mm). It is observed from the images that evacuation started approximately midway through the powder bed for both LH206 and SV010 (Fig 2(c) left and right). Despite the variation in the powder morphology that occurs over time due to fluctuations in the gas phase, which could have implications on the transient nature of the drug delivery, it is important to note that the final evacuation time value (as shown in section 3.3 onwards) is just as repeatable with a grid as it is without one. The evacuation time value with a grid is also significantly shorter (see section 3.3 onwards). Of further interest to note here is that using a grid enables the complete evacuation of powder out of the pocket (not shown in Fig. 2) compared to cases with no grid and enables evacuation at lower flow-rates. The reader should note that the images shown here are line-integrated and therefore, for the turbulent cases, the powder interface is the average location of the interface across the entire pocket, which is also partly responsible for the “overlap” of the powder-air interface lines halfway through the acquisition time shown in Fig. 2c.

Figure 2.

Powder evacuation morphology with and without using grid

3.2. Effect of powder properties and flow rate on evacuation behaviour

Prior to a more detailed analysis of the powder evacuation behaviour, the effect of powder properties on evacuation from the pocket is examined for the 5×5 channel at different flow rates (40, 60, 80, and 100 L/min) without a grid as benchmark cases. These flow rates correspond to Reynolds numbers that range from 9023 to 22560, which are all within the turbulent flow regime. Figure 3 shows the evacuation percentage (number of pixels occupied by powder/total number of pixels in image) plotted vs. time for a selection of cases. The evacuation percentage graphs shown in Fig. 3 indicate that at lower flow rate the powder properties have a significant effect on the evacuation behaviour of powder from the pocket and show an evacuation rate much slower than the higher flow-rate case. At 40 L/min, SV010 (powder with the lowest Carr’s index and highest mass median diameter) has the fastest evacuation time, compared to LH206 and SV003. The highest evacuation percentage is also observed for SV010, which indicates that its properties are conducive to a more effective emptying of the powder pocket compared to the other two cases. At higher flow rate (100L/min), it can be seen that the evacuation curves for the three powders nearly overlap, further confirming the diminishing role of powder properties at higher flow-rates, which gradually occurs as aerodynamic forces dominate the fluidisation process, both due to increased mean flow but also turbulence (Islam and Cleary, 2012). LH206 has the slowest evacuation time in all the tested flow rates. Of further interest is that the powder evacuation percentage curves increase monotonically at 100 L/min, while for 40 L/min, the gradient is not as uniform, indicating that equilibrium between aerodynamic forces and inter-particle forces between powder particle have likely not been achieved, again highlighting the important role of powder properties at low flow-rates towards the emptying of a pocket.

Figure 3.

A) Powder evacuation percentage (top) and B) powder volume flow rate (bottom) vs. time for 40 L/min (left) and 100 L/min (right) for a range of powders

The powder volume flow-rate is calculated by subtracting the number of black pixels (corresponding to evacuated powder) for each frame from the number of black pixels at the start of evacuation. The result from the subtraction operation is multiplied by a scale factor to obtain a value in mm², which is multiplied by the channel width to approximate the evacuated volume for each frame. Whilst this is a limitation of the technique, it is enough to obtain general information evacuation characteristics. The evacuated volume is divided by the time at which each frame was captured to calculate an equivalent powder volume flow rate. Figure 4 plots the normalized powder volume flow-rate vs. normalized time providing a quantitative marker for the real volume of powder that vacates from the pocket. The normalized time is the real-time divided by the time it takes for the flow to travel across the pocket (transit time) and the powder volume-flowrate is normalized by the air-flow rate. This enables different flow-rate curves to collapse when aerodynamic forces are the key driving force behind the fluidization process, hence allowing the discussion to focus on characteristics independent from increasing the airflow velocity. From the volume flow rate curves, it can be seen that for all cases, the powder evacuation is an unsteady flow process especially at the start of evacuation where the powder volume flow rate increases sharply, followed by a decrease until it reaches a steady state value. At 40 L/min, the difference between peak volume flow rate and the volume flow at the end of evacuation is smaller compared to 100 L/min demonstrating a less severe transient process which occurs at lower air flow velocities. What can be seen for cases at 60, 80 and 100 L/min is that all of the curves of powder volume flow-rate collapse over a similar range of air flow-rates and over a similar non-dimensional time interval. This interesting result suggests that the increase in the shear force from the flow in the channel is a key driver at flow-rates 60 L/min and above, which are highly turbulent cases. In contrast for 40 L/min, it can be seen that the curves show distinct evacuation behaviour which deviate from the higher flow-rate cases. This is likely once again related to the interparticle forces between the powder particles that become more apparent at a lower flow-rate, in this case 40 L/min which is a common physiological flow-rate. As breathing is naturally transient, these results also indicate that it would be expected to see a time-variant dependence on carrier properties during inspiration, and this merits further study in practical devices and under flow-rates which are typical of individuals with very inhibited breathing.

Figure 4.

Normalized powder volume flow-rate (powder flow-rate/air flow rate) plotted vs. normalized time (real-time/transit time) for a range of powders and flow-rates

3.3. Effect of grid on powder evacuation

In this section, the effect of using an upstream grid on the powder fluidization characteristics is investigated. Results will initially focus on the grid with a blockage ratio of ~30%. The grid was placed at different distances (x) before the powder pocket. Both sieved power (SV010) and milled powder (LH206) were used to study the effects of the grid at 80 and 100 L/min.

Observing Fig.5, it can be seen that adding a grid at a location of x~0-15mm from the powder bed can significantly alter the initial evacuation characteristics of the powder, making the overall evacuation rate faster, as evident through an increased gradient of the evacuation percentage curve. The most significant effect on the initial evacuation rate occurs for grids placed at a distance of x = 10mm or closer which equates to an x/M~5. From the powder volume flow rate curves (Fig. 5-right), it can be seen that the peak of volume flow rate decreases with increasing the grid distance (X). This powder volume flow rate is entirely unsteady for X = 0mm which is demonstrated by the fact that a) it does not reach a steady-state value even until the very end of the evacuation process and b) the lack of a plateau being observed in evacuation % curve (Fig. 5 left). This is in contrast to what is observed with larger grid distance (x > 10mm), where a much longer steady-state evacuation period is visible and the time of evacuation is similar to the cases without a grid. The addition of a grid therefore clearly makes evacuation more rapid and is also less likely to result in residual powder in the pocket.

Figure 5.

Effect of the location of a grid with blockage ratio of ~30% for 10X10 for SV010 at 100L/min on: (a) evacuation percentage vs. time, and (b) powder volume flow rate vs. time

Whilst the grid clearly accelerates the powder emptying process it is important to consider the effect of increased resistance, which from a patient perspective would result in a requirement for a higher pressure drop to achieve a particular flow-rate. The role of device resistance has been recently studied (Clark et al., 2019; Weers and Clark, 2017), and there is evidence showing that patients can receive adequate dosage even with high resistance DPIs. Nevertheless, further work should be undertaken to examine the change in resistance due to addition of the types of grids shown in this study.

3.3.1. Effect of blockage ratio

The effect of changing the blockage ratio of the grid on the time-history of evacuation and powder-flow rate is summarized in Fig. 6 for powder SV010 at 100 L/min for grids with blockage ratios of ~25%, ~30%, and ~40%. The grids were placed at the powder pocket edge (x=0). It can be seen that the grid with the low blockage ratio (σ= 25%) has a faster evacuation time compared to the other tested grids. For grids with blockage ratio of ~30% and ~40%, the powder volume flow rate is nearly the same. Whilst it is clear that the blockage ratio does influence the overall evacuation time (suggesting a different flow-field over the powder-to be confirmed in section 3.4), it has a very minimal effect on the final evacuation percentage (residual powder). Further analysis of evacuation behaviour is therefore now restricted to the mid-range (~30%) blockage ratio grid only.

Figure 6.

The effect of blockage ratio of the grid on evacuation % vs. time (left) and powder volume flow-rate vs. time (right) for SV010 at 100L/min in 10X10 channel

3.3.2. Summary of grid effects on overall evacuation time and final evacuation percentage

Figure 7 summarizes the effect of a grid on the evacuation characteristics of powders SV010 and LH206 for 100 L/min (Fig. 7 left) and 80 L/min (Fig. 7 right). The dashed lines pertain to the evacuation time (left vertical axes) and the solid lines to the final evacuation % (right vertical axes). The evacuation time here has been calculated from data similar to Fig. 5 (left) by finding the time at which the evacuation percentage reaches its maximum value and remains constant with time. The evacuation % is the maximum evacuation % reached from data similar to Fig. 5 (left).

Figure 7.

Evacuation time and evacuation percentage vs. grid position (x) for grid with blockage ratio of ~30% for 10X10 channel: (a) 100 L/min and (b) 80 L/min. (Note: Evacuation time (evacuation percentage) for no grid cases for LH206 and SV010 are 35s (57%) and 16s (74%) respectively at 80 L/min, and 15.3s (85.5%) and 10.5s (95%), respectively at 100 L/min

Figure 7 shows that the powder evacuation percentage decreases with increasing grid distance (X) for LH206 and SV010 at both 80 L/min and 100 L/min, and evacuation time increases consistently with increasing distance (X). Of particular interest to note is that the effect of mixing in the flow (controlled through positioning of the grid) affects the evacuation time for both SV010 and LH206 in a very similar way, and this is clearly the case for both 80 and 100 L/min. This agrees with measurements using a different laser extinction technique which indicated that at higher levels of turbulence, the evacuation time was largely insensitive to powder properties (Mahmoudi et al., 2019). However, results from this study suggest that the final evacuation % is dependent on powder properties. Of further interest to note here is that while positioning a grid very close to the powder bed has very clear effects on both evacuation time and final evacuation %, moving the grid past a particular point can result in the opposite effect. For instance, Fig. 7 (right) shows that moving the grid past x=10mm can result in a final evacuation % lower than the benchmark case of no-grid, and likewise the evacuation time can also become slower. The reasons for this are not fully understood but may be related to more fully developed turbulence (closer to isotropic decaying conditions (Pope, 2001)) which could influence local mixing, hence increasing deposition due to the higher residence time of the powder particles, however further work is required to fully substantiate this claim.

The slope of decrease in evacuation percentage with grid distance for LH206 is steeper compared to SV010 at 80 and 100 L/min, potentially due to the mass median diameter of LH206 (84.9 μm) which is smaller compared to SV010 (112 μm). The smaller particle sizes of LH206 and higher fine particle percentages are likely to lead to a higher bulk inter-particle force and therefore lower final evacuation percentages (an indicator of the residue left behind in the pocket).

3.4. Effect of grid on powder flow field

Figure 8 presents the mean velocity magnitude contours of the powder emerging out of the pocket in the area just above the pocket, for all four powders, at 100 L/min. These contours are shown for the grid with the ~30% blockage ratio. The powder velocity magnitude has its lowest value just adjacent to the grid and increases as the particles travel downstream. This is expected due to the Stokes number being greater than unity for all of these particles. The velocity magnitude of particles nearer to the lower wall is seen to be lower than those particles which migrate to the top of the wall. SV003 has the highest velocity magnitude because it has the lowest mass median diameter (x50=61.6 μm) and the two milled powders LH200 and LH206 have lower velocity magnitudes as their mass median diameter is 74.5 μm, and 84.9 μm respectively. The lowest velocity range is for SV010 which has the largest mass median diameter of 112 μm, showing physical consistency in the results.

Figure 8.

Mean velocity magnitude contours (above powder pocket) overlayed with vector field for a grid with blockage ratio of ~30% at x=0 mm for all four powders at 100 L/min in the 10X10 channel. For each contour plot, the left edge is flush with the edge of the powder pocket

The vector fields indicate that whilst the powder properties can control the overall range of the mean velocity, the general structure of the flow is largely unchanged, thereby suggesting that the powder properties, other than modulating the overall magnitude of the velocity, will not otherwise significantly influence the mean flow-field in the vicinity of the powder bed. This would support results of evacuation times and previous laser extinction results (Mahmoudi et al., 2019), which generally point to a diminishing role of powder properties at high levels of turbulence.

More useful in terms of characterisation of the grid behaviour is measurement of the turbulence intensity, as a measure of how significant the fluctuating component in the flow is. The key role of the turbulence generating grid is to increase local mixing in the vicinity of the pocket, and this has clearly been demonstrated through the influence on evacuation times and powder flow-rates presented earlier. Figure 6 demonstrated that the lowest blockage ratio grid can lead to an improved overall evacuation time, suggesting that careful grid design can lead to control over the fluidization behaviour. Figure 9 shows measurements of the turbulence intensity $\left({}^{u_{\mathit{rms}}^{\prime }}∕_{\stackrel{‒}{u}}\right)$ from two powders (LH200 and SV003) for the three different blockage ratio grids at two locations downstream of the grid. The turbulence intensity $\left({}^{u_{\mathit{rms}}^{\prime }}∕_{\stackrel{‒}{u}}\right)$ was calculated for each point in the domain by taking the standard deviation of the time series of streamwise velocities and normalizing by the time-averaged velocity over a time interval where the mean velocity was near constant. The standard deviation of the instantaneous velocity is equivalent to the root mean square of velocity fluctuations (as defined in the context of Reynolds averaging). The reader should note that this “turbulence intensity” is not strictly an intensity as measured in the gas phase. The quantity is a measure of the turbulent fluctuations as seen by the carrier particles themselves many of which have Stokes numbers greater than unity-and hence velocity measurements here are dependent on the particle size.

Figure 9.

Profiles of turbulence intensity in x-direction, just above powder pocket for blockage ratios of ~25%, ~30 and ~40% (X=0mm) at a flow-rate of 100 L/min.

The centreline streamwise turbulence intensity $\left({}^{u_{\mathit{rms}}^{\prime }}∕_{\stackrel{‒}{u}}\right)$ for both LH200 and SV003 clearly increases after using a grid with a blockage ratio of ~25% compared to ~30%. This supports an argument that the improved evacuation time for the 25% blockage ratio grid (Fig. 6) is due to enhanced mixing. No obvious conclusions can be reached by comparing the two powders, however the results do indicate that PIV measurements taken using LH200 particles result in a higher “turbulence intensity”. The turbulence intensity range for the point located furthest from the channel walls was (0.25 to 0.45) for LH200 compared to (0.13 to 0.3) for SV003, which is likely related to the higher percentage of fines in LH200 (resulting in greater fluctuations associated with smaller particles). It is also noticed that the local turbulence intensity decreases with an increase in distance from the grid related to the decreased effect of mixing moving further away from the grid bars.

Conclusion and Limitations

The effect of an upstream grid on the fluidization of lactose carrier powders has been examined using high speed imaging and particle image velocimetry. To the best of the authors’ knowledge this presents one of the only generalized studies on the influence of grid generated velocity fluctuation on the fluidization of pharmaceutical powders. Use of a grid placed just upstream of a powder bed reduces the evacuation time, can significantly reduce residual powder left in the pocket, and also enable evacuation of powder at lower mean flow-rates. The final point may be of importance for individuals with lower inspiratory capacity.

Consistent with previous work, the powder properties partly control the evacuation behaviour at lower flow-rates, with the role of properties becoming diminished at higher degrees of turbulent mixing due to the dominance of aerodynamic forces. With the grids studied here, a grid with low blockage ratio can reduce the evacuation time, and PIV results indicate that it is due to a higher local turbulence intensity. Therefore, use of such grids can be considered as an additional design tool for controlling the behaviour of DPI flows, however further work is required to ascertain this over a wider range of grid designs and flows, and to assess the effect of an upstream grid on deagglomeration for carrier and drug blends as well as for pure drug powders, e.g. Mannitol. Future work should focus on a wider range of grid designs as well as microscopic PIV measurements to enable a particle size sub-ranged measurement of velocity, as well as investigation with carrier and active drug mixtures.