DPI Flow Visualization: Initial Experimental Results

1
DPI Flow Visualization Initial Experimental
Results

• Rob Tuley
• John Shrimpton

2
Introduction

• Motivation
• Experimental Simplifications
• Experimental Design Procedure
• Some Results
• Evacuation Timescales
• Scope and Limitations
• Conclusions
• Future work

3
Motivation

• To gain a better understanding of how and why the
  Diskus pocket evacuation occurs.
• Ultimate aim is to computationally simulate this
  process this is the first exploratory set of
  experiments to produce a data-set to validate
  these simulations.

4
Simplifications - Inhalation

• Inhalation simulated by a linear pressure ramp.

5
Simplifications - Geometry

• We tested 4 different geometries, although the
  results from only two are being presented today

Square U-bend Diskus Replica
fully developed turbulent in-flow approx real-scale pocket geometry exact. includes cross-hairs at in-flow and out-flow.
6
Experiment Design

• Experimental rig designed in a modular fashion
  for added flexibility.

log pressure sensor data
inhalator
pocket geometry
power supply and vacuum pressure source.
optical setup
7
Geometry modules
8
Pocket filling

• Excess powder swiped off from surface. Low
  compression.

9
Inhalator module

• Pressure is controlled through a PI control
  feedback loop.

10
Optical Setup

• The square U-bend geometry is backlit with a
  halogen 1kW floodlight.
• Diskus replica is front lit with the same halogen
  flood.

11
Complete Setup
12
Parameter Space

• The two variables examined during the course of
  the experiments were
• Pressure ramp gradient 4 different ramps were
  examined, both steeper and shallower than the
  average real inhalation ramp of -30kPa/s.
• Powder type 4 powders were chosen with various
  shape and cohesive properties

time
Pulse -30kPa/s -10kPa/s -3.33kPa/s
pressure
13
Powders
Glass particles 0-50mm Nearly perfectly spherical. Small cohesion forces.
Aluminium particles 0-44mm Non-spherical flakes. Small cohesion forces.
Lactose 6 fines blend Non-spherical. Medium cohesion forces.
Lactose 15.8 fines blend Non-spherical. Large cohesion forces.
14
Video Results
channel
channel
Straight U-bend
Diskus Replica
pocket
pocket
15
Square U-bend flow features
Transient blow-back effect
16
Diskus Replica flow features
Strong influence of cross-hairs
17
Key Points

• No visible difference in evacuation between glass
  and aluminium particles.
• Two modes of evacuation EROSION
  (glass/aluminium) and FRACTURE (15.8 fines
  lactose). Lactose 6.5 fines behaviour lies
  between these two extremes.
• Differences evident between the two geometries
  considered, but type of pocket evacuation and
  powder behaviour visibly similar.
• Difficult to quantify effect of pressure ramps

18
Intensity Post-processing
Normalised intensity 1
Area processed
Normalised intensity 0
19
Evacuation Timescales
20
Evacuation Pressures
21
Pressure Ramps - Glass
22
Pressure Ramps - Glass
23
Pressure Ramps - Lactose
24
Pressure Ramps - Lactose
25
Key Points

• Aluminium and glass evacuations are quantifiably
  equivalent.
• Lactose 15.8 fines blend seem to evacuate based
  on instantaneous pressure, not gradient.
• Glass/Aluminium evacuations, and to a lesser
  extent lactose 6 fines blend seem to be
  influenced by pressure gradient, but this may be
  due to hidden 3D effects.

26
Scope

• Various pocket geometries can be connected to the
  simulated inhalation apparatus.
• Labview software PI control feedback loop means
  any pressure profile can be simulated.
• Running the software on a faster PC and DAQ card
  would give better control loop performance.
• Options to use various optical setups/different
  lighting, etc.

27
Current Limitations

• There can be a certain amount of delay in the
  pressure control PI loop.
• With the current setup there is a high
  noise-to-signal ratio from the pressure sensors.
• Backlighting provides a 2D visualization
  possibility that certain 3D effects might be
  hidden.
• Difficult to quantify possible leakage effects.
• Pocket filling method could be improved.
• Repeatability

28
Repeatability

• Each parameter combination was repeated a total
  of three times.
• Some parameter combinations were repeated after a
  gap of a few days and changed geometries.
• Examine glass and lactose 15.8 fines, at time
  0.09s and 0.06s respectively, -30kPa/s.

29
Repeatability (glass 0-50mm)
30
Repeatability (lactose 15.8 fines)
31
Conclusions

• The evacuation of the simplified 2D U-shaped
  geometry is similar to the evacuation from the 3D
  Diskus replica geometry.
• Cohesion has a much more important effect on the
  pocket evacuation than shape.
• The evacuation of lactose is based on
  instantaneous pressure, not on pressure gradient.
  This conclusion cannot be extended to any of the
  other particle types.
• A clearer understanding has been achieved of the
  timescales and pressures required for particle
  evacuation.

32
Future Work

• Using more sophisticated camera equipment from
  EPSRC, better resolution images possible (approx
  1000 x 1000 pixels).
• This batch of experimental results not analysed
  fully quantitatively at present, but possibility
  of using techniques such as PIV or variance maps
  on future results.
• Experiment with a wider range of lactose blends?
• Data to be used on a validation basis for
  computational simulations.

33
Discussion

• Data CD with a copy of this presentation and
  further experiment description and results
  available from Mark Palmer.

Questions?
34
Extras
35
Extras
36
Extras
37
Extras