190
D. Markauskas and H. Kruggel-Emden
slightly more particles remain on the screening surface between t = 2.5–10 s, but
afterwards the results fit very well. At an amplitude of A = 0.8 mm (see Fig. 22b),
the obtained fraction retained values are higher when water is added. The numerical
and experimental results for a smaller amount of water (M = 5%) fit together very
well. In the DEM simulations, however, the fraction retained for a larger amount of
water (M = 10%) is slightly overpredicted. In the configuration with smaller particle diameters d 1/ 2/ 3 = 3/5/7 mm, the influence of liquid is more pronounced (see
Fig. 22c). For smaller particle sizes used in this investigation, the capillary forces
become larger relative to the weight force, which is relatively small because of the
low density of POM. The fraction retained is similar after t = 20 s for M = 0% and
M = 5% and only slightly larger for M = 10% due to the pegging of particles in the
dry case. The DEM simulations reveal the same trends but show some deviations
between t = 1–10 s.
When analyzing the initial configuration with glass spheres (see Fig. 22d), a
larger amount of water increases the fraction retained obtained experimentally and
numerically. However, the influence of the liquid is relatively low due to the large
particle masses. The simulation results show some deviations between t = 2–5 s
under the influence of water. After that, the results fit very well. The same trends
can be seen with an amplitude of A = 0.8 mm (see Fig. 22e). However, here all
results obtained are closer to each other. The results derived using glass spheres with
smaller particle diameters of d 1/ 2/ 3 = 3/5/7 mm (comp. Fig. 22f) differ greatly from
those obtained using POM spheres. In particular, the experimental results are close
to each other with slightly larger values when more water is added. Because of the
larger density of glass particles, the influence of the capillary force is smaller than
for POM. The simulation results show a bit more differences and slightly overpredict
the fraction retained until t ≈ 7 s and underpredict it afterwards. Due to the pegging
of the dry particles, less particles remain on the screening surface at t = 20 s if liquid
is added. Overall, the simulation results agree well with the experimental ones.
4.3.3 Benchmarking of Extended Screening Models
Based on the successful validation of moist DEM screening simulations in Sect. 4.3.2,
in this section, results on the three in [123] introduced and in Sect. 3.2.4 outlined
modified process models for batch screening under the influence of moisture are
shown when adjusting their parameters to fit the results obtained by DEM simulations
such as outlined in Sect. 4.3.2 which are further extended. To compare the introduced
models over a larger number of investigations, an average deviation of the fraction
retained per particle size class obtained by DEM simulations and process models is
calculated. For the different undersized particle classes i, the average of the obtained
fractional deviations is given by
l
i=1
j
k=1 |Y mod (i, k) − Y sim (i, k)|
/( j · r ),
where j is the total number of considered time steps k and r is the total number of
undersized fractions (here r = 2). The total time of the screening process t = 20 s is
divided into intervals of Δt = 0.5 s here.
D. Markauskas and H. Kruggel-Emden
slightly more particles remain on the screening surface between t = 2.5–10 s, but
afterwards the results fit very well. At an amplitude of A = 0.8 mm (see Fig. 22b),
the obtained fraction retained values are higher when water is added. The numerical
and experimental results for a smaller amount of water (M = 5%) fit together very
well. In the DEM simulations, however, the fraction retained for a larger amount of
water (M = 10%) is slightly overpredicted. In the configuration with smaller particle diameters d 1/ 2/ 3 = 3/5/7 mm, the influence of liquid is more pronounced (see
Fig. 22c). For smaller particle sizes used in this investigation, the capillary forces
become larger relative to the weight force, which is relatively small because of the
low density of POM. The fraction retained is similar after t = 20 s for M = 0% and
M = 5% and only slightly larger for M = 10% due to the pegging of particles in the
dry case. The DEM simulations reveal the same trends but show some deviations
between t = 1–10 s.
When analyzing the initial configuration with glass spheres (see Fig. 22d), a
larger amount of water increases the fraction retained obtained experimentally and
numerically. However, the influence of the liquid is relatively low due to the large
particle masses. The simulation results show some deviations between t = 2–5 s
under the influence of water. After that, the results fit very well. The same trends
can be seen with an amplitude of A = 0.8 mm (see Fig. 22e). However, here all
results obtained are closer to each other. The results derived using glass spheres with
smaller particle diameters of d 1/ 2/ 3 = 3/5/7 mm (comp. Fig. 22f) differ greatly from
those obtained using POM spheres. In particular, the experimental results are close
to each other with slightly larger values when more water is added. Because of the
larger density of glass particles, the influence of the capillary force is smaller than
for POM. The simulation results show a bit more differences and slightly overpredict
the fraction retained until t ≈ 7 s and underpredict it afterwards. Due to the pegging
of the dry particles, less particles remain on the screening surface at t = 20 s if liquid
is added. Overall, the simulation results agree well with the experimental ones.
4.3.3 Benchmarking of Extended Screening Models
Based on the successful validation of moist DEM screening simulations in Sect. 4.3.2,
in this section, results on the three in [123] introduced and in Sect. 3.2.4 outlined
modified process models for batch screening under the influence of moisture are
shown when adjusting their parameters to fit the results obtained by DEM simulations
such as outlined in Sect. 4.3.2 which are further extended. To compare the introduced
models over a larger number of investigations, an average deviation of the fraction
retained per particle size class obtained by DEM simulations and process models is
calculated. For the different undersized particle classes i, the average of the obtained
fractional deviations is given by
l
i=1
j
k=1 |Y mod (i, k) − Y sim (i, k)|
/( j · r ),
where j is the total number of considered time steps k and r is the total number of
undersized fractions (here r = 2). The total time of the screening process t = 20 s is
divided into intervals of Δt = 0.5 s here.