240
M. Gleiss and H. Nirschl
sediment transport [5]. For the scale-up of solid bowl centrifuges the manufacturers
use numbers of experiments on a pilot scale. This procedure is time-consuming and
cost-intensive and does not allow any prediction about the dynamic process behavior
[6]. For a theoretical description of the transient response of solid bowl centrifuges, it
is necessary to consider flow conditions and separation behavior. A major challenge
in depicting the separation process in solid bowl centrifuges arises from the fact,
that particle separation depends on the residence time in the apparatus. In contrast
to thickeners, the flow direction in solid bowl centrifuges results in a classification
of particles along the rotor [7].
In the field of flowsheet simulation, it is important to predict the steady-state
or dynamic behavior of a process plant using time-efficient mathematical models.
For other applications such as Model Predictive Control (MPC) it is essential to
calculate faster than real time to enable a coupling of dynamic modeling with the
process control level. Mesh-based methods such as Computational Fluid Dynamics
(CFD) are not suitable for flowsheet simulations [8, 9]. CFD simulations rather
serve to derive parameters, that are not achievable experimentally from numerical
experiments [10, 11].
The following section begins with an overview of the experimental setup to investigate material functions for the settling behavior and the cake formation process.
The dynamic model for solid bowl centrifuges uses material functions for processorientated dynamic simulations. Subsequently, different experimental methods to
characterize the system and residence time behavior for decanter centrifuges are
presented. Based on the investigations of material and process behavior the following section deals with the mathematical modeling of the dynamic behavior of
continuously working decanter centrifuges. The comparison of dynamic simulations
with pilot-scale experiments for decanter centrifuges shows the applicability of the
developed numerical approach. The following section shows the development of a
dynamic model for semi-continuously operating tubular centrifuges. Simulations of
a tubular centrifuge on a pilot scale reveal a different process behavior of tubular
centrifuges compared to decanter centrifuges. Finally, the conclusion summarizes
the main results and gives a short outlook on further work.
2 Material Functions and Separation Properties
The properties of the disperse and the fluid phase such as particle size, particle
shape, solid volume fraction, physicochemical properties, density of solid and liquid
as well as dynamic viscosity have a significant influence on the material behavior
during mechanical fluid separation [12]. Due to the large number of influencing
quantities, there are no generally applicable models for the arbitrary product. Rather,
it is preferable to investigate the material properties in a laboratory apparatus and to
develop material functions for the theoretical description of the separation process
[13]. The use of well-established laboratory equipment, such as beaker centrifuges
or filters is one way to achieve this goal [14, 15]. In the case of solid bowl centrifuges,
M. Gleiss and H. Nirschl
sediment transport [5]. For the scale-up of solid bowl centrifuges the manufacturers
use numbers of experiments on a pilot scale. This procedure is time-consuming and
cost-intensive and does not allow any prediction about the dynamic process behavior
[6]. For a theoretical description of the transient response of solid bowl centrifuges, it
is necessary to consider flow conditions and separation behavior. A major challenge
in depicting the separation process in solid bowl centrifuges arises from the fact,
that particle separation depends on the residence time in the apparatus. In contrast
to thickeners, the flow direction in solid bowl centrifuges results in a classification
of particles along the rotor [7].
In the field of flowsheet simulation, it is important to predict the steady-state
or dynamic behavior of a process plant using time-efficient mathematical models.
For other applications such as Model Predictive Control (MPC) it is essential to
calculate faster than real time to enable a coupling of dynamic modeling with the
process control level. Mesh-based methods such as Computational Fluid Dynamics
(CFD) are not suitable for flowsheet simulations [8, 9]. CFD simulations rather
serve to derive parameters, that are not achievable experimentally from numerical
experiments [10, 11].
The following section begins with an overview of the experimental setup to investigate material functions for the settling behavior and the cake formation process.
The dynamic model for solid bowl centrifuges uses material functions for processorientated dynamic simulations. Subsequently, different experimental methods to
characterize the system and residence time behavior for decanter centrifuges are
presented. Based on the investigations of material and process behavior the following section deals with the mathematical modeling of the dynamic behavior of
continuously working decanter centrifuges. The comparison of dynamic simulations
with pilot-scale experiments for decanter centrifuges shows the applicability of the
developed numerical approach. The following section shows the development of a
dynamic model for semi-continuously operating tubular centrifuges. Simulations of
a tubular centrifuge on a pilot scale reveal a different process behavior of tubular
centrifuges compared to decanter centrifuges. Finally, the conclusion summarizes
the main results and gives a short outlook on further work.
2 Material Functions and Separation Properties
The properties of the disperse and the fluid phase such as particle size, particle
shape, solid volume fraction, physicochemical properties, density of solid and liquid
as well as dynamic viscosity have a significant influence on the material behavior
during mechanical fluid separation [12]. Due to the large number of influencing
quantities, there are no generally applicable models for the arbitrary product. Rather,
it is preferable to investigate the material properties in a laboratory apparatus and to
develop material functions for the theoretical description of the separation process
[13]. The use of well-established laboratory equipment, such as beaker centrifuges
or filters is one way to achieve this goal [14, 15]. In the case of solid bowl centrifuges,