Powder dilation as indicator for cohesivity and flowability

Using the model powders as in A. M. Faqih, Chaudhuri, Alexander et al. (2006) and applying discrete element simulations, A. M. Faqih, Chaudhuri, Muzzio et al. (2006) investigated powder dilation induced by powder flow in the GDR. The behaviour of powders at 50% fill level and 7 RPM, 16 RPM, and 29 RPM was recorded with a video camera and video footage was analyzed with a pixel-counting computer program to determine the relative volume of powder to bed void. Equation 6.2 was used to calculate powder dilation; VInitial was the initial volume of powder bed

after the powder in the GDR was shaken horizontally and vertically for an unreported fixed number of times and allowed to settle under its own weight, and VNew was the volume of powder

measured at the first 11 revolutions.

Powder dilation=VNewVInitial

V_Initial 100% (6.2)

It was observed that radical increase in dilation occurred between 1–2 revolutions before the powders reached a steady state (A. M. Faqih, Chaudhuri, Muzzio, et al., 2006). Dilation increased with increasing cohesiveness, and an excellent correlation between Flow Index and powder dilation (Eq. 6.2) was demonstrated. Drum speeds caused short-term variability in dilation regardless of the powders used, and the speed effect generally became insignificant once steady state was reached. For cohesive powders, such as the ~60 μm microcrystalline cellulose and ~50 μm milled lactose powders, higher drum speeds led to a higher frequency of avalanches without significantly changing the dynamics of flow. For the more free-flowing ~100 μm spherical lactose powder, dilation increased with increasing rotation rate at steady state; continuous powder flow, avalanches that were neither discrete nor separated, and aperiodic cascading of the bed surface were seen. At higher rotation rates, the powder became airborne at

the top of the flowing layer. With regards to the effect of cylinder diameter (A. M. Faqih, Chaudhuri, Muzzio, et al., 2006), dilation decreased monotonically when the diameter of the cylinder was between ~50 mm and ~200 mm, and a sharp rise was observed when the diameter increased from ~200 mm to ~254 mm. The monotonic decrease was attributed to less constrained flow and increase in shear that caused powder agglomerates to break up, allowing more powders to move as individual particles. In the ~254 mm drum, although powder flowed freely, avalanches would cease before reaching the end wall of the cylinder. The authors stated that shear was insufficient to break up powder agglomerates and allow the powders to cascade down the cylinder; hence powder dilation remained high (A. M. Faqih, Chaudhuri, Muzzio, et al., 2006).

Results from discrete element simulations were consistent with experimental data (Alexander, et al., 2006; A. M. Faqih, Chaudhuri, Muzzio, et al., 2006). It was further noted by A. M. Faqih, Chaudhuri, Muzzio et al. (2006) that air entrainment was a consequence of powder dilation at low drum rotations and not a root cause because the influence of interstitial air was neglected in the simulations.

6.2.5 Summary of literature review

Powder tumbling is common in the handling and processing of powder systems. The GDR has been built and developed to measure the avalanche activity of powders that move under their own weight when rotated in a cylindrical drum; manipulation of the avalanche data gives information on powder flowability. With limited experimental data from selected powder systems, the GDR has shown promise as a robust tool for flowability characterization; the utility of the GDR has included the identification of powder flow regimes at different drum speeds, identification of transitions between flow regimes, and comparison of avalanche activity of powders from different Geldart Groups. Two flow indices, namely the Flow Index, Equation 6.1, and powder dilation, Equation 6.2, have been proposed as indicators to powder flowability and cohesivity; there has been consistency between experimental results and computer simulations.

6.3 Aims

1. To measure the avalanche activity of samples of milled lactose powders, sand, refractory dust, and glass beads.

2. To investigate the influence of drum fill level and d*32 on the avalanche activity of the

selected powders.

3. To investigate the relationships between avalanche activity, cohesion measured by shear testing, and powder bed collapse measured by fluidization for the selected powders. 4. To assess the relationship between avalanche activity and Geldart Powder Classification.

6.4 Materials, Methods and Analysis 6.4.1 Materials

The powders used were samples of milled lactose LP4, LM1 and LP1, sand S1, refractory dust RD1, and glass beads B8; Table 6.1 gives information on the d*32, span, tc/Hmf, C0, and Geldart

Classification (Geldart, 1973) for the powders. LP4 was a Group C powder, LM1, S1, and RD1 were Group A powders, and LP1 was a A/B boundary powder. B8 was a Group B powder; its d32

was measured by sieve analysis.

Table 6.1 Information on d*32, span, tc/Hmf, C0, and Geldart Classification for lactose LP4,

sand S1, refractory dust RD1, lactose LM1, lactose LP1, and glass beads B8

Powders d*32 [μm] Span [-] tc/Hmf [s m1_{] C0 [kPa] Geldart Classification}

LP4 28.9 2.26 0.0534 C S1 40.1 1.43 64.20 0.0427 A RD1 41.5 1.57 35.47 0.0632 A LM1 58.0 2.23 0.0527 A LP1 150.8 1.75 0.0245 A/B B8 193.0 0.15 B 6.4.2 Experimental setup

Figure 6.2 is a front view photo of the GDR. The cylindrical drum was made of Perspex. One end of the drum was permanently closed by a Perspex disc and the other end was fitted with a snug- fit Perspex lid; the internal diameter and length were respectively 192.5 mm and 292.5 mm. The GDR included a drive system that consisted of a motor, a drive pulley, a pivot plate, four support rollers on which the drum was mounted, a built-in load cell positioned at the drum centre line and under the pivot plate, and a control panel that comprised, from left to right, an on/off switch, a drum directional control switch, a dial that controlled the drum speed, and a L.E.D. drum speed indicator. The drive pulley transferred power from the motor and drove the rotation of the support rollers and drum. The pivot plate held the drum atop the load cell, and the position of the pivot plate – lifted up or touching the load cell – was controlled by a lever. The load cell was connected to a data acquisition system and a computer equipped with custom software. Schematic diagrams of a similar GDR are given in Davies et al. (2002).

Figure 6.2 A photo of the Gravitational Displacement Rheometer 6.4.3 Setup and operation of GDR

The setting up and operation of the GDR were based on the operator’s guide by Pingali and Kick (2013); the operator’s guide contains information on setting up of the GDR, understanding the

Graphical User Interface in the custom software, and step-by-step instructions on operating the

GDR, which included imaging setup, load cell calibration, powder installation, and Flow Index experimentation.

Drum rotation rates were measured by visual observation and with a stopwatch; the number displayed on the L.E.D. drum speed indicator was used as a guide.

Circularity, irregularity, and dimensional tolerances of the cylindrical drum, which could cause wobble, were checked prior to experiments. The empty drum was rotated between ~5 RPM and ~30 RPM, and the load cell signal at each incremental drum speed was recorded at a sampling rate of 30 Hz; 1,800 data points were collected and the standard deviation of drum weight shift, ws, at each incremental drum speed was determined. The plot of ws data versus

drum speed was shown in Figure A6.1 in Appendix 6.2; ws increased with increasing drum

speed, see the filled diamond data points with the label Initial drum, and ws ranged from 0.0223–

0.0315 kg, indicating wobble caused by the cylindrical drum. A new cylindrical drum of the same dimensions was made and its wobble was checked. The ws data were relatively constant

within the range of drum speeds used, see Figure A6.1; the values of ws were 0.0204–0.0247 kg

in Run 1, 0.0217–0.0240 kg in Run 2, 0.0179–0.0218 kg in Run 3, and 0.0211–0.0252 kg in Run 4. The new drum gave significant improvement and was subsequently used for experiments.

Avalanche activity was measured at 5 RPM, 10 RPM, 15 RPM, 20 RPM, 25 RPM, and 30 RPM; 1,800 data points were sampled at 30 Hz, ws was determined, and Flow Index was

calculated following Equation 6.1. All powder samples were initialized before the experiments; this was done by rotating the powders at 15 RPM for ~2 minutes prior to avalanche activity measurement. In the investigation on the effect of drum fill level, the fill levels were set at 20%, 30%, 40%, and 50% on a volume basis; loose poured bulk density measured by the modified

NZS3111 method, 0,mNZS3111, see Section 4.4.2 and Table 4.1 in Chapter 4, was used to calculate

the sample mass required for each fill level. An infrared video camera that was connected to the computer and controlled by the custom software was used to capture random images of powder bed surface during tumbling; the ‘Print Screen’ function was used as the function of the video camera was limited to powder dilation measurement and not video recording. All experiments were repeated once or twice, and the average ws was calculated.

6.4.4 Analysis

Plots of standard deviation of drum weight shift, ws, versus drum speed were constructed.

Interpretation of the data was done taking into consideration d*32, C0, Geldart Powder

Classification, and powder bed collapse and tc/Hmf; see respectively Section 2.5.1 of Chapter 2,

Section 3.5.3 of Chapter 3, Section 5.2.2 of Chapter 5, and Section 5.5.3 of Chapter 5. 6.5 Results