Methodology
Process variables investigated in the experimental design
| Factor |
Low level (-) |
Middle level (0) |
High level (+) |
| Atomizing air pressure (bar) |
0.3 |
1.2 |
2.0 |
| Column velocity (m/s) |
5 |
7 |
9 |
| Insert diameter (mm) |
24 |
30 |
40 |
Response variables for evaluating the granulations:
- LOD after liquid addition, fraction of discharged yield > 2 mm, granule size distribution and friability (on unmilled granulations)
- Poured bulk and tapped densities, Hausner ratio, Carr index (on milled granulations)
Table 6 - Experimental design*** for investigating the influence of 4 selected PG process variables
| Run |
Atomizing air pressure (bar) |
Column velocity (m/s)** |
Insert diameter (mm) |
Air cap area/opening |
| 1 |
- |
+ |
0 |
+ |
| 2 |
+ |
+ |
- |
0 |
| 3 |
- |
+ |
+ |
- |
| 4 |
- |
- |
+ |
- |
| 5 |
0 |
0 |
- |
+ |
| 6 |
- |
- |
- |
+ |
| 7 |
- |
+ |
+ |
0 |
| 8 |
+ |
- |
+ |
- |
| 9 |
+ |
- |
0 |
+ |
| 10 |
- |
- |
- |
0 |
| 11 |
+ |
+ |
- |
- |
| 12 |
+ |
+ |
0 |
+ |
| 13 |
+ |
- |
0 |
0 |
| 2a* |
+ |
+ |
- |
0 |
| 6a* |
- |
- |
- |
+ |
| 4a* |
- |
- |
+ |
- |
| 2b* |
+ |
+ |
- |
0 |
| 1a* |
- |
+ |
0 |
+ |
- * Indicates repeat runs. Data collected included in statistical analysis.
- ** Corresponding to column velocities at high, middle and low levels, binder spray rates set at 36, 28 and 20 g/min and air flow volume rates at 143, 111 and 80 m3/h, respectively. Spray rates and airflow volume rates chosen for maintaining the same change in humidity between inlet and outlet conditions during binder liquid addition stage.
Table 7 - Physical properties of lactose 200M powder mix for the experimental design study
| Property |
Value |
| Mass median diameter, D50%* |
0.032mm |
| Span* |
1.48 |
| Poured bulk density |
0.547 (± 0.008) g/mL |
| Tapped density |
0.808 (± 0.004) g/mL |
| Hausner ratio |
1.48 (± 0.03) |
| Carr index |
32.33 (± 1.33) % |
| LOD |
1.91 (± 0.46) % |
- * Particle size analysis by laser scattering (LA-910, Horiba)
Results
Experimental design study - Granule characteristics
Table 8 - LOD after liquid addition, discharged yield and fraction of yield > 2 mm
| Run |
LOD after liquid addition (%) |
Discharged yield (%) |
Fraction of discharged yield > 2 mm (%) |
| 1 |
9.34 |
89.43 |
0.23 |
| 2 |
5.42 |
89.01 |
0.34 |
| 3 |
10.47 |
93.93 |
2.73 |
| 4 |
2.18 |
48.25 |
13.77 |
| 5 |
8.39 |
93.32 |
0.23 |
| 6 |
5.80 |
80.52 |
17.01 |
| 7 |
9.33 |
93.00 |
0.11 |
| 8 |
9.05 |
88.70 |
2.04 |
| 9 |
8.97 |
90.27 |
0.02 |
| 10 |
10.16 |
87.85 |
0.15 |
| 11 |
10.25 |
85.62 |
2.61 |
| 12 |
4.81 |
91.48 |
0.15 |
| 13 |
9.07 |
90.23 |
0.06 |
| 2a* |
8.74 |
81.21 |
1.55 |
| 6a* |
6.06 |
70.50 |
14.72 |
| 4a* |
2.61 |
41.74 |
12.88 |
| 2b* |
4.06 |
84.45 |
1.85 |
| 1a* |
11.37 |
88.28 |
0.39 |
- *Indicates repeat runs.
- Low LODs (about 2%) for runs 4 and 4a - due to non-homogeneous distribution of binder solution to powder mass. Over-wetting in column resulted in formation of caked material around the spray nozzle. While the caked material was over-wetted, material fluidizing outside the column (where sampling for LOD was performed) was under-wetted.
- Most PG runs had discharged yields of 70-90% and gave granulations with < 3% w/w oversize particles (> 2 mm). Low discharged yields for runs 4 and 4a (42-48%) and relatively high proportions of oversize particles for runs 4, 4a, 6 and 6a (12-17%) - attributed to higher tendency for over-wetting in column with combination of low column velocity (associated with low airflow volume) and low atomizing air pressure.
Table 9 - Granule size distribution and friability
| Run |
D50%(mm) |
D25%(mm) |
D75%(mm) |
Span |
Friability (%) |
| 1 |
0.390 |
0.563 |
0.252 |
0.80 |
2.08 |
| 2 |
0.189 |
0.269 |
0.128 |
0.75 |
3.87 |
| 3 |
0.600 |
0.935 |
0.373 |
0.94 |
1.26 |
| 4 |
0.732 |
1.329 |
0.353 |
1.33 |
7.75 |
| 5 |
0.232 |
0.388 |
0.156 |
1.00 |
3.65 |
| 6 |
0.382 |
0.706 |
0.230 |
1.25 |
3.75 |
| 7 |
0.335 |
0.487 |
0.222 |
0.79 |
2.77 |
| 8 |
0.468 |
0.831 |
0.292 |
1.15 |
4.20 |
| 9 |
0.263 |
0.390 |
0.201 |
0.72 |
5.72 |
| 10 |
0.286 |
0.445 |
0.210 |
0.82 |
4.24 |
| 11 |
0.239 |
0.454 |
0.153 |
1.26 |
3.41 |
| 12 |
0.196 |
0.269 |
0.134 |
0.69 |
5.90 |
| 13 |
0.254 |
0.386 |
0.195 |
0.75 |
5.93 |
| 2a* |
0.210 |
0.331 |
0.141 |
0.90 |
9.34 |
| 6a* |
0.384 |
0.631 |
0.240 |
1.02 |
5.31 |
| 4a* |
0.805 |
1.412 |
0.447 |
1.20 |
8.35 |
| 2b* |
0.227 |
0.367 |
0.146 |
0.97 |
5.89 |
| 1a* |
0.368 |
0.559 |
0.233 |
0.91 |
1.80 |
Table 10 - Poured bulk and tapped densities, Hausner ratio and Car rindex values of milled PG granules
| Run |
Poured bulk density (g/mL) |
Tapped density (g/mL) |
Hausner ratio |
Carr Index (%) |
| 1 |
0.524 |
0.616 |
1.18 |
14.98 |
| 2 |
0.515 |
0.612 |
1.19 |
15.86 |
| 3 |
0.527 |
0.614 |
1.17 |
14.27 |
| 4 |
0.505 |
0.597 |
1.18 |
15.34 |
| 5 |
0.491 |
0.589 |
1.20 |
16.65 |
| 6 |
0.478 |
0.572 |
1.20 |
16.49 |
| 7 |
0.479 |
0.571 |
1.19 |
16.11 |
| 8 |
0.456 |
0.532 |
1.17 |
14.29 |
| 9 |
0.435 |
0.525 |
1.21 |
17.20 |
| 10 |
0.462 |
0.554 |
1.20 |
16.60 |
| 11 |
0.546 |
0.649 |
1.19 |
15.97 |
| 12 |
0.491 |
0.590 |
1.20 |
16.73 |
| 13 |
0.437 |
0.528 |
1.21 |
17.18 |
| 2a* |
0.529 |
0.629 |
1.19 |
16.00 |
| 6a* |
0.476 |
0.566 |
1.19 |
15.85 |
| 4a* |
0.498 |
0.582 |
1.17 |
14.47 |
| 2b* |
0.520 |
0.626 |
1.20 |
16.98 |
| 1a* |
0.531 |
0.620 |
1.17 |
14.41 |
- *Indicates repeat runs.
- Poured bulk and tapped densities of milled granulations were lower than those of the original lactose 200M powder mix.
- Hausner ratio and Carr index values were also lower, indicating that the milled PG granules had better flow properties than the powder mix.
Statistical analysis
Table 11 - P values from ANOVA and R2 values from summary of Fit data
| Responses for granule characterization |
Significance of process variables in rangesunder investigation at 95% confidence interval |
R2 |
| LOD after binder liquid addition (%) |
Not significant (P = 0.1774) |
0.5029 |
| Fraction of discharged yield > 2 mm (%) |
Not significant (P = 0.1226) |
0.5448 |
| Granule friability (%) |
Not significant (P = 0.0467, i.e. 0.05) |
0.6335 |
| D50% |
Significant (P <0.0001) |
0.9272 |
- LOD after liquid addition, fraction of discharged yield >2 mm and granule friability responses were independent of the process variables.
- Granule size distribution and density responses were dependent on the process variables.
Table 12 - Effects and interactions of the process variables on granule responses
| Response |
Process variables |
P value |
Effect/interaction |
| D50% |
Atomizing air pressure
Insert diameter X Column velocity
Insert diameter
Column velocity
Air cap area/opening
Air cap area/opening X Atomizing air pressure |
*<0.0001
*0.0058
*0.0151
**0.0637
**0.0986
0.2251 |
-0.235
-0.160
0.143
-0.063
-0.087
0.054 |
| Poured bulk density |
Column velocity
Atomizing air pressure
Insert diameter X Column velocity
Insert diameter
Air cap area/opening
Air cap area/opening X Atomizing air pressure |
*<0.0001
*0.0001
*0.0011
*0.0048
*0.0154
0.7933 |
0.057
-0.041
-0.038
-0.032
-0.026
0.002 |
| Tapped density |
Column velocity
Atomizing air pressure
Insert diameter X Column velocity
Insert diameter
Air cap area/opening
Air cap area/opening X Atomizing air pressure |
*<0.0001
*0.0002
*0.0005
*0.0013
*0.0130
0.3246 |
0.063
-0.042
-0.047
-0.039
-0.028
0.008 |
- * Statistically significant (P< 0.05)
- ** 0.05 <P< 0.1
- For mean granule size, atomizing airpressure was the most important process variable while air cap area/opening was the least important.
- Column velocity had the most important effect on density responses of milled granulations while air cap area/opening was the least critical process variable.
- An interaction between column velocity and insert diameter was significant at P <0.05.
- Granule size increased with increase in insert diameter and decrease in atomizing air pressure, column velocity and air cap area/opening (Figure 7).
- Poured bulk and tapped densities of the milled granulations generally increased following an increase in column velocity and a decrease in atomizing air pressure, insert diameter and air cap area/opening (Figure 7).