(505b) Feeder Discharge Prediction Using Discrete Element Method (DEM) Simulations

Feeder discharge
prediction using Discrete Element Method (DEM) simulations

Martina Trogrlic¹, Peter Toson¹, Eva
Siegmann¹, Dalibor Jajcevic¹, Johannes Khinast¹

&

Pankaj Doshi², Daniel Blackwood², Mary
T. am Ende²

¹
Research Center Pharmaceutical Engineering, Inffeldgasse 13, 8010 Graz, Austria

²
Pfizer, Inc., Worldwide Research and Development, Groton, CT, USA

Keywords: Pharmaceutical Manufacturing; Solids Flow,
Handling, and Processing; Equipment

The
use of powder feeder is one of the essential processes enabling continuous
manufacturing in the pharmaceutical, chemical, food and other industries. In
the drug production process, the ability to feed the correct amount of active
pharmaceutical ingredient (API) and excipients can make the difference between good
and poor product quality. Challenges with a constant feed rate emerge from the
material properties of the powder, where optimal feeder parameter settings for
one powder often fails for a different one with e.g. different material
properties like inter-particle cohesion level and/or particle size distribution
(PSD).  The feed rate is also influenced by the packing density of the powder
in the hopper, which depends on the powder state of consolidation, environmental
factors, and events such as refilling[1].
 Significant effort has investigated ways of achieving a consistent feed factor
(mass fed per screw rotation), since the variance in the feed rate is often
transmitted downstream to the next unit operation[2].

As
shown in Fig.1, a twin-screw feeder consists of a hopper, impeller, barrel, two
screws and a screen plate at the end. The role of the impeller is to break
clumps of the powder and to prevent bridging. The screen at the end is intended
to break up any agglomerates formed as the powder progresses along the length
of the screw.

DEM
method provides insight into feeder discharge time, feed rate, screw torque and
solid fraction in the screw. Another advantage of using DEM simulations is the
ability to predict the feeding behavior for a range of powders that are
available in scarce amounts or are too costly to run full hopper experiments.

The
number and the size of particles are key factors that influence duration of the
simulation[3].
Particle number in the case of full hopper reaches on the order of several
million, and in combination with the rotating geometries, the computation time
can become significantly long. For simulating the complete discharge of the
feeder in Fig.1 with calibrated cohesive material, it would take over 5 months
of calculation time, for approximately 20 minutes of real process time. In
order to reduce the computational time, instead of running one discharge simulation
of full hopper, multiple piecewise simulations were run where a part of the
powder bed was replaced with a newly developed “pressure boundary condition”. Since
solid fraction in the screw and consequently the feed rate changes directly
with the hopper fill level, different fill levels were modeled by changing the
force acting on the upper boundary surface. The resulting feed rate of the
piecewise simulations was compared with the feed rate of the full hopper
discharge, each modeled with linear spring-dashpot (LSDP) contact model, and
very good agreement was demonstrated. In this way the computational time for
feeder discharge simulation was reduced to approximately 1 month, resulting in
computational savings of 4 months.

Virtual
design of experiments (vDOEs) with different screw and impeller speed, screen
presence, hopper fill level and calibrated materials have been performed.  The
results reveal very good agreement for the feed rate. The difference between
cases with and without an end screen shows that the screen ensures a constant
feed rate during discharge process. The particle contact model and material
parameter affect the powder solid fraction inside the screw, and thus also the
feed rate. DEM simulations with a cohesive force model are in better agreement with
experimental results than the classic non-cohesive linear spring-dashpot model.
This shows that for capturing the flow behavior of the powder, material
calibration is of essential importance.

Fig. 1 Feeder with
cut-away powder bed inside hopper to reveal the impeller feeding the screw
region.

[1]         W. E. Engisch and F. J. Muzzio, “Feedrate deviations
caused by hopper refill of loss-in-weight feeders,” Powder Technol.,
vol. 283, pp. 389–400, 2015.

[2]         R. Weinekötter and L.
Reh, “Continuous Mixing of Fine Particles,” Part. Part. Syst. Charact.,
vol. 12, no. 1, pp. 46–53, 1995.

[3]         D. S. Nasato, C.
Goniva, S. Pirker, and C. Kloss, “Coarse graining for large-scale DEM
simulations of particle flow - An investigation on contact and cohesion
models,” Procedia Eng., vol. 102, pp. 1484–1490, 2015.