|
CAPE-OPEN UPDATE, Volume 8
CAPE-OPEN UPDATE is a publication of the CAPE-OPEN Laboratories Network (CO-LaN), a non-profit consortium for the development of the CAPE-OPEN standard.
STAFF LISTING:
Kerry Irons, Editor
Editorial Board: Peter Banks, Bertrand Braunschweig, Celeste Colantonio, Ronald-Alexander Klein, Werner Merk, Hans Pingen, Michel Pons
Technical Support: ADDUCE GmbH
Flowsheeting for Solids Processes
Ernst-Ulrich Hartge (
hartge@tu-harburg.de
)
Joachim Werther (
werther@tu-harburg.de
)
Guenter Gruhn (
gruhn@tu-harburg.de
)
Jens Schmidt (
jsschmidt@dow.com
)
Matthias Pododda (
pogodda@tuhh.de
)
Claus Reimers (
claus.reimers@tuhh.de
)
Daniel Schwier (
schwier@tuhh.de
)
. Chemical Engineering I / III
Technical University Hamburg-Harburg, D-21071 Hamburg, Germany
(
http://www.tu-harburg.de/)
While flow sheet simulation is a common method to design and optimize chemical processes involving only
fluids, in solids processing it is still common practice to design one apparatus separately from the other. There is a lack of suitable
methods to combine single apparatus models with a process model and to treat this process as a whole. The present paper will elaborate
the reasons for this situation and will present a concept and a program package under development, which will allow the application of
methods of flow sheet simulation on complete processes, in which at least one major component is a solid.
Introduction
Today, flow sheeting program packages are commonly used in chemical engineering for the design of processes involving fluids. The
simulation of processes which involve solids or solids and fluids is not as advanced, and it is common practice to design and optimize
each apparatus separately from the others, neglecting its influence and dependence from the neighbouring processes. This allows
finding optimal operating conditions for each unit operation or apparatus. But such a sequential optimization of individual units
will not always lead to a global optimum for the whole process.
With respect to flow sheet simulation there are some major differences between fluid processes and solids processes, resulting from
the dispersed nature of solids:
While a fluid can be characterized by a limited number of concentrated parameters as temperature, pressure, composition and some
compound specific parameters, a lot more information is required to describe solid particles, e.g. size, porosity, humidity, shape.
Most of the additional parameters are distributed parameters, e.g. the particle size distribution. Even the composition of
solid particles is often not homogeneous but has to be treated as a distributed parameter. And, of course a distributed composition of
the solids will cause most parameters which depend on the composition of the solids to be distributed, too.
In many cases a certain property may be dependent from another distributed property, e.g. during a drying process the water content of
coarse particles will be higher than that of small particles, which dry much faster than the coarse ones. But even particles of the same
size will not have under any circumstances the same humidity: if there is a residence time distribution there will be a distribution
of humidity within each particle size class. This leads to a hierarchy of dependent distributed properties.
In many cases the exact composition of solids is not known (and often not of interest) as for example in the case of coal or sand.
In these cases the solids have to be defined as their own species. This approach, however, makes it nearly impossible to obtain or
calculate exact properties for this species without doing experiments.
Even when the properties of single particles are known, generally applicable methods to calculate the bulk properties from the
properties of the individual particles are often missing.
The performance of an apparatus is often significantly influenced by the geometry and can not always be scaled easily; e.g.
the diameter of a hydrocyclone is decisive for the separation process, influencing both the cut size and the separation sharpness.
The above mentioned differences require a numerical treatment of solids which strongly differs from the treatment of fluid components.
Thus it is hardly possible to extend existing simulation packages for fluid processes to the simulation of solids processes without
restructuring and re-implementing it from scratch. Distributed parameters require additional mathematical treatment, e.g. the
fulfillment of population balances, which add new families of equations to the complex mathematics of flow sheeting systems and thus
need new or adapted solvers.
For these reasons it was decided at the Hamburg University of Technology to put efforts into the development of a completely new system
for the simulation of solids processes. These efforts led to a project together with 10 of the leading academic groups in particle
technology from different German universities with the aim to develop a simulation system for solids processes called SolidSim.
In addition, many industrial users of solids processes and suppliers for solids processing equipment contribute with their expertise to
this development. This two year project is now funded by the AiF (Arbeitsgemeinschaft industrieller Forschungsvereinigungen) within
their ZUTECH program.
Objectives for the Development of SolidSim
The main objectives for the development of SolidSim as a flow sheet simulator for solids processes can be derived from the special
requirements of solids processes as given above. Thus, the simulator should provide a stream structure, which allows a description of
solids with distributed parameters, allowing also for a hierarchy of dependent properties. Furthermore mathematical methods and solvers
must be provided which allow the treatment of population balances.
Even though the development is focused on solids processes, SolidSim should not be limited to them, but should allow also the simulation
of related fluid processes. Since it is far beyond the scope of the present project to develop all model required for such a simulation
by itself, an important requirement is to follow open standards which allow the inclusion of external modules into SolidSim or of
SolidSim into other simulation environments. To achieve these aims it was decided to follow the CAPE-OPEN standards wherever possible
and extend them in order to include the treatment of the distributed parameters for solids.
Another objective was to develop an open system such that new modules can easily be added, that different models for the same
apparatus can be made available at the same time and that modules can easily be replaced by improved modules without affecting the
whole system.
Besides these objectives concerning the inner functionality of the system there was furthermore the requirement to develop a user
friendly system with an ergonomic easy-to-use graphical user interface (GUI), to allow engineers to build process following their view.
Modelling solids with distributed parameters and fluids significantly increases the heterogeneity of the mathematical equations to be
solved. Since solving these equations requires quite different solvers, it was decided not to use an equation oriented simulation
approach but to build a block-oriented simulation system.
As software platform the Microsoft Windows® operating system family has been chosen.
Structure of SoldiSim
The program system was divided into three major parts: the simulation environment, the stream objects, and the model library with the
unit models. The simulation environment provides the basis of the system. It provides the graphical user interface, some basic
functionality as reporting, printing, saving and loading data, and it controls the calculation sequence. The material stream object
provides the structures necessary to transport the information on the fluids and solids from one module to the next one, and to provide
the modules with all properties of the materials needed by the modules. The Unit Model simulates a single apparatus or unit operation.
These basic elements are connected by different interfaces as defined by the CAPE-OPEN-Standard. However, some interfaces had to be
extended within this project for the treatment of solids parameters.
Additionally, communication with other simulation packages is also possible using the CAPE-OPEN interfaces. A detail flow sheet
simulated with SolidSim may be used as one module in an external CAPE-OPEN compliant simulation environment, using standard interfaces
for the communication. Another method is to incorporate modules from other packages as unit models inside the SolidSim environment or
to use external property packages inside SolidSim.
Model Library
The model library is basically a collection of independent software modules containing one or several models for a certain apparatus or
a unit operation. In the framework of the current project, a basic library with models for the most important solids process steps will
be implemented. For each apparatus, models of different complexity and different requirements with respect to the input data will be
used. This allows the use of different models in different stages of the design process, e.g. quite simple models which need only a few,
easy to measure or to estimate data in an early design phase and a more complex model with higher data quality requirements to simulate
or optimize an existing plant.
The software modules are implemented as COM objects, so that they can added and removed during runtime without influencing the remaining
simulation system.
The modules communicate with the simulation environment using the CAPE-OPEN ICapeUnit… interfaces and with the connected
streams for the handling of distributed parameters extended ICapeThermo+… interfaces.
The ICapeUnit… interfaces are used to control the program logic and execution sequence, the ICapeThermo+
… interfaces are used to query all information about the incoming materials. In the case of solids processes, additional information e.g. on the geometry of the apparatus is usually necessary; such information as well as model parameters have to be requested from the user by a user interface provided by each module.
Usually only parameters purely related to the apparatus should be queried this way, but for some models, parameters related to the
combination of apparatus geometry and solids properties are needed and also have to be queried this way.
In order to facilitate the development of a unit model, a ‘base unit class’ has been developed, which forms the basis of all
currently developed modules. The task of this base unit is twofold, first it will hide all implementation details for the communication
between the different software components from the "engineering oriented" developer of the unit model. The second task is to
provide and ensure a minimal functionality, e.g. to copy input to output streams, to check mass balances, to provide methods for
connecting streams to the ports of the apparatus, etc. In the simplest case the model implementer should only have to implement specific
calculation and initialisation routines and to provide some basic information about entrance and exit ports.
Within the present project, the model library will be far from being completed, but for the most important areas of solids processing,
models will be implemented. Below you can find a list of solids processing steps for which models will be developed together with the
name of the responsible project partner.
Separation, classification, fluidisation, framework; Werther, Hamburg
Simulation Environment; Gruhn, Hamburg
Crystallisation, dissolving; Kind, Karlsruhe
Agglomeration, granulation; Mörl, Magdeburg
Crushing in mills; Peukert, Erlangen
Filtration of fluids; Ripperger, Dresden
Gas filtration; E. Schmidt, Wuppertal
Separation processes in centrifuges; Stahl, Karlsruhe
Convection drying; Tsotsas, Magdeburg
Liquid sprayers; Walzel, Dortmund
Dosing and conveying; Wirth, Erlangen
Stream Objects
One major development task in the project is the development of the stream objects, i.e. the structures which store the current status
of a material stream connecting two subsequent models or units in the flowsheet and which have to provide the models with all properties
of the incoming materials, i.e. fluids and solids. The material stream object mainly provides access to a compound list, a property
package providing the thermo-physical properties of the compounds, a list of global constants and a phase manager.
While existing approaches could be used for the implementation of the fluid part and for the storage of concentrated properties, the
implementation of the solids part with the distributed properties had to be newly developed.
The design goal was a structure which allows the efficient storage of and access to complex hierarchies of nested and dependent
distributed solids properties. In addition, the stream object should be able to provide the information to the unit models in an
appropriate manner, i.e. it should only give the information needed by the module, hiding all additional complexity.
The problem with the dependent distributed parameters will be clarified with the following example in which a gas cyclone is used to
separate solids from a gas. The gas cyclone separates solids from a gas due to the centrifugal forces inside the vortex.
Thus, the separation is governed by the terminal velocity of the particle in a centrifugal field. A further influence on the separation
efficiency is the solids loading µ defined as solids mass flow divided by the gas mass flow. An increasing solids loading at the
cyclone entrance increases the total separation efficiency and changes the fractional separation efficiency.
Now let us take a solids stream made up from two different kinds of solids with given particle size distributions and given densities.
The simplest approach to calculate the separation efficiency would be to calculate the separation for the two compounds in sequence, but
this not possible due to the interaction of the solids and due to the influence of the total solids loading at the entrance. The
approach which is used here is to give the cyclone model the information in a manner as it is needed, i.e. to give the total gas flow,
the total solids flow and a distribution of the terminal velocities of the solids, not distinguishing between the two different kinds
of solids. Then the cyclone model gives back to the stream object the total solids separation efficiency and a movement matrix, i.e.
the information about which mass fraction in each terminal velocity class is separated from the gas flow and which mass fraction
could not be separated. The stream object then calculates from the movement matrix the separation for each particle size of each
compound. Thus, the fact that the entrance flow was formed of different compounds is hidden from the cyclone model, which allows
using a standard separation model, which is usually designed for only a single compound.
Similarly the separation of solids with a dependent distribution of humidity or any other property can be done by the standard gas
cyclone model, as long as the stream object is able to calculate the influence of the dependent property on the distribution of terminal
velocities. Similar scenarios could be found for most of the solids processing steps.
In order to facilitate the calculations which have to be performed by the stream object, the properties for the solids are stored in
an n-dimensional matrix with n being the total number of different properties under consideration. In this way the dependences between
the properties are depicted without implying any hierarchy.
Status of the Simulation Package SolidSim
Currently all the basic structures as described above are implemented and first versions of the unit models have been implemented.
Testing of the simulation system is ongoing using different processes we are kindly provided information on by industrial partners.
One of the test cases simulated is a plant for the separation and dewatering of contaminated sewage sludge. The aim of this process is
to maximize the amount of sand with low contamination and to minimize the amount of highly contaminated fines under the restriction that a certain allowed level of contamination of the sand is not exceeded.
The flow sheet of this process includes several unit models, for example sieves, a hydrocyclone, decanter, elutriator and thickener.
For further testing we strive to simulate the multi stage evaporation of a caustic soda solution as occurring within the caustic soda
process. Here the crystallization, including recycling of crystallization product, has to be calculated showing the ability of SolidSim
not only to deal with pure solids processes but to be able to handle solid-fluid interactions as well.
For further information please contact Prof. J. Werther at werther@tuhh.de
| 
