7 Dynamic Simulation of Mechanical Fluid Separation in Solid …
265
Fig. 20 Dynamic simulation
of grade efficiency dependent
on time and volumetric flow
rate. Simulation setup is C =
19200 and φ in = 0.005 [20]
result the sediment volume can be derived for each time step. Figure 21 presents the
sediment height as a function of the rotor length for C = 9600, C = 28900 and C
= 38500. Here, the g-force refers to the bowl radius. Each iso-line corresponds to a
constant time. The interval between two lines is t = 10 min. The results indicate
deviating process behavior for the sediment build-up as a function of the volume
flow. A thin sediment layer is formed in the rotor for C = 9600 at time t = 90 min.
For C = 38,500 the rotor is almost completely filled. This process behavior results
from the increasing amount of the separated solid for a higher g-force.
6 Conclusion
This work presents the development of two dynamic models for continuous working
decanter centrifuges and semi-batch tubular centrifuges. Both models are computationally efficient and therefore suitable for dynamic flowsheet simulation and Model
Predictive Control. Dynamic simulation requires material functions and the residence time behavior to predict the real process behavior with sufficient accuracy. In
contrast to the -theory, which carries out experiments on a pilot scale, the settling
behavior and the sediment build-up were investigated sing well-established laboratory equipment. This allows a detailed numerical investigation of the process level
for solid-bowl centrifuges.
Another important parameter for the description of continuous and semi-batch
machines are the flow conditions, which influence the residence time of the particles.
For this reason, three methods have been applied to investigate the flow conditions:
the experimental residence time measurement, CFD simulations and characterization
of the system behavior. The experimental data for the system behavior results from
a stepwise change of particle size distribution and solids volume fraction at the feed
265
Fig. 20 Dynamic simulation
of grade efficiency dependent
on time and volumetric flow
rate. Simulation setup is C =
19200 and φ in = 0.005 [20]
result the sediment volume can be derived for each time step. Figure 21 presents the
sediment height as a function of the rotor length for C = 9600, C = 28900 and C
= 38500. Here, the g-force refers to the bowl radius. Each iso-line corresponds to a
constant time. The interval between two lines is t = 10 min. The results indicate
deviating process behavior for the sediment build-up as a function of the volume
flow. A thin sediment layer is formed in the rotor for C = 9600 at time t = 90 min.
For C = 38,500 the rotor is almost completely filled. This process behavior results
from the increasing amount of the separated solid for a higher g-force.
6 Conclusion
This work presents the development of two dynamic models for continuous working
decanter centrifuges and semi-batch tubular centrifuges. Both models are computationally efficient and therefore suitable for dynamic flowsheet simulation and Model
Predictive Control. Dynamic simulation requires material functions and the residence time behavior to predict the real process behavior with sufficient accuracy. In
contrast to the -theory, which carries out experiments on a pilot scale, the settling
behavior and the sediment build-up were investigated sing well-established laboratory equipment. This allows a detailed numerical investigation of the process level
for solid-bowl centrifuges.
Another important parameter for the description of continuous and semi-batch
machines are the flow conditions, which influence the residence time of the particles.
For this reason, three methods have been applied to investigate the flow conditions:
the experimental residence time measurement, CFD simulations and characterization
of the system behavior. The experimental data for the system behavior results from
a stepwise change of particle size distribution and solids volume fraction at the feed