CHAPTER ONE

INTRODUCTION

The computer aided framework presented in this thesis can ease the job of the process engineer
(or engineers) actually carrying out the preliminary design task on request from management.
The task the engineers are performing can be described as the qualified selection among the
many options for designing the process. The idea is not that this can be done in a completely
automatic manner with framework. Rather, the framework will assist, through user
interactions, in generating designs and after that, in screening among the alternatives so that
only the feasible candidates are considered in the final evaluation. Many activities in the area
of process development are labeled process synthesis. In the content of the term, process
synthesis refers to the generation of a preliminary design for primarily chemical and
pharmaceutical manufacturing process.

Also, a framework has been provided in building blocks for the integrated framework as
described and detailed on an algorithm which was developed as part of this thesis for the
design of distillation columns. In doing this, it also represents a detailed algorithm for the
generation of process flowsheets and superstructures. To overcome this modeling bottleneck a
systemization of modeling as well as the development of advanced computer aided modeling is
environments is required. Several research groups have focused on the development of such
tools. The requirements are demanding and can only be fulfilled in the longer run. This is not
only due to the significant implementation effort required, but also due to the need of further
investigations of the fundamentals of computer aided flowsheeting. In order to explore at least
a number of these issues, a research program has been set up with the objective to design,
implement and evaluate the advanced modeling.

MODELLING

Models are not just equations. In addition to the equations, model representations should
include model assumptions and limitations, information can be on specifications of degrees of
freedom and on model initialization etc. Furthermore, all decisions taken during a modeling
project needs to be recorded to render the modeling process transparent. The development and
storage of the same process need to be supported. As a consequence, the versions of a model
which have been built during a modeling project for whatever purpose need to be documented
together with their interrelationships. It shows them how engineers have problems in
formulating process models by writing modeler and the modeling tool must be moved from the

1
equation level to the knowledge level. This does not only allows user interaction based on
chemical engineering concepts every engineer is familiar with but also forms a basis for the set
up of correct and reusable models and modeling tools should adopt, store, retrieve modeling
know-how to be used to guide the process of model development.

Another point of view is that a repository of predefined model building blocks of fine
granularity like equations describing reaction kinetics or heat and mass transfer must be
provided to support the modification(for the same process after modification or for another
similar process). Automation of parts of the modeling process is an explorative and creative-
activity, automation of parts of the modeling process is possible. This includes knowledge
propagation and documentation as well as report generation.

2
 CHAPTER TWO
FLOW SHEETS

Diagram of figures of flow sheets are used in many areas of science for different applications
for example, to illustrate the flow of data in computer science. In this flow sheets are used to
represent manufacturing process, typically chemical or pharmaceutical processes.

The basic objects in a flow sheet are units (units operations) represented by symbols as
shown in the diagram and corrections ( stream flows) represented by lines usually with arrows
indicating the flow direction.

Example of unit symbols representing unit operations (equipment).

The connections can either connect two units, thereby indicating a mass flow or energy
flow or both between the units. if only connected in either their end points or starting point, an
inlet stream or an outlet stream, respectively, is indicated. A dashed line is often used to
indicate information flow in the flow sheet. Furthermore, a connection is implicitly
representing a state. The definition of state includes quantitative intensive and extensive
properties i.e, temperature, pressure, high content and mass flow. The state, however. Is not
limited to quantitative information but can also be used to includes qualitative information as
proposed by papaletandri and pistikoponlons ( 1960). The qualitative information usually
includes the name of the components that are present in the stream, but may also includes, for

3
example, a heat exchange label ( hot or cold stream). If desired, the state associated to the
stream can be highlighted by a circle.

Two different representations of a connection, to the right the state associated to the stream is
indicated by the number ‘3’ which can then be used for reference for example in a table.

A major part of this computer aided tools and flow sheeting (or super structure) generation
which involves more than just creating a flow diagram consisting of stream and unit object
rather the tools assist in selecting the objects (equipment) for the flow sheet and for deriving
the specification for these objects.

Representation of superstructures

A flow sheet super structure is an extended flow sheet where compared to conventional flow
sheets, the extension is the ability to represent different flow sheet alternatives for the process
in the same figure considering for a example the case where for a given process the best
selection has to be made among two different reactors. By the rise of logical splitters exist
only in the context of the superstructure model not physically in the process., in the model the
splitter have one or more integer (usually binary) variable assigned to them these variables are
used to indicate which connection from the splitter is part of the current flow sheet when a
value is assigned to all the logical splitter variable of the superstructure, the super structure
model represents one (current) flow sheet when a value is assigned to all the logical splitter
variable of the super structure, the super structure model represents one (current) flow sheet
when a value is assigned to all the logical splitter variables of the superstructure, the super
structure model represents one (current) flow sheet the principle of the logical splitter is used
in the super structure with two reactor described above problem exist grass root designed
(where the process is designed from scratch) band retrofit design. In the second type an
existing process flow sheet has to be changed in order to match new objectives in some cases
the operation of the existing flow sheet is changed to satisfy the new objectives while the rest
remains fixed.

4
 In the flow sheet synthesis problem the input and output is known the rest is to be determined by

a) Mathematical or optimization techniques

b) Thermodynamic/physical insight

This corresponds also to the division of the reported literature in sections. While application
of typical frame combination do not seek to report optimal flow sheets mathematical
(structure optimization) techniques that claim to determine option flow sheet are limited by
the availability an application range of the model and or the superstructure. If methodologies
of the three types can be integrated into one frame-work their combined use allows
exploitation of the strengths of each methodology while compensating for the weakness for
example, designed flow sheet which is a result from a type which is not necessarily optimal

SIMULATION AND OPTIMIZATION

Simulation of unit operations and process flow sheets has been carried out ever since
computer became available to chemical engineers one might argue that calculations were
performed even before computer became available, but all practical process simulation
problems are solved with the aid of computer today. Simulation of process flow sheet
simulation can be performed with either a steady-state approach (calculating a steady
operation point of the process) or a dynamic approach may be used. In the former the
calculation of each unit and the task of converging the overall balances are solved
simultaneously with one common solver.

5
OPTIMIZATION

The problem solved with optimization can be described as a measure of what is the best
solution needed. Therefore an objective function is defined for the problem, usually a
mathematical expression related to the yearly cost or profit for the process the outcome of
solving an optimization problem is the optimal value for certain (decision) variables some of
them (dependent variables) may however be bounded to lie within the defined constraints of
the problem a general mathematical form of the process optimization problems can be
written as:

Min f (X)

s.t

g(F)<=0

h(X)=0

where X is a vector of continuous variables (the decision and dependent variable), f (F) is the
objective function , g (X) and h (X) represents vector of inequality and equality.

ALGORITHMS AND TOOLS

Structural optimization tool

Once a flow sheet superstructure with equipment designed parameter gas been obtained with
the integrated frame work the optimal solution can be obtained by applying structure
optimization techniques structural optimization techniques have not been developed as part
of this thesis , however two different algorithms have been applied for the solution of case
studies in this thesis. The first approach has been developed by kravanja and grossman
(1994), while the second, called a interactive MINLP solver’’ which is the main algorithm
used in this thesis was developed by Henriksen etal (2000).

6
 CHAPTER THREE
ALGORITHMS AND TOOLS

Structural optimization tool

Once a flow sheet superstructure with equipment designed parameter gas been obtained
with the integrated frame work the optimal solution can be obtained by applying
structure optimization techniques structural optimization techniques have not been
developed as part of this thesis , however two different algorithms have been applied for
the solution of case studies in this thesis. The first approach has been developed by
kravanja and grossman (1994), while the second, called a interactive MINLP solver’’
which is the main algorithm used in this thesis was developed by Henriksen etal (2000).

The main advantage of this interactive algorithm is that it has been implemented in the
integrated computer aided system(ICAS) where also the rest of ther framework
developed in this thesis have been incorporated. This way “Interactive MLNLP
SOLVER” can act as part of integrated framework. The algorithm is based on the
decomposition approached, where a NNP sub-problem(the prional problem ) is solved it
the inner loop and a master problem is solved in the outer loop. The solver can operate in
two modes.

a. |Equation system mode- Here all equations are given explicitly. This is the typical
way of using NLNLP solver.
b. Simulation mode- In this operation mode, process model equations can solved by
process simulation. This way the process model equations does not need to be
supplied explicitly by the user.

7
8
 CHAPTER FOUR
APPLICATION EXAMPLES

The application of the framework is illustrated through several illustrative examples in this
chapter. Some of the examples focuses on the specific part of the framework. (Eg Estimation of
kinetics). In other to highlight the corresponding features of that part. Other examples aim at
illustrating the general idea of integration which is the central to this design. By applying
several part of the framework in an integrated manner. The main focus for each example is
summarized below.

REACTION KINECTICS

Reaction Analysis tool has been used for this estimation of the reaction kinetics. These
complexity is presented starting with hydrolysis of ethylene oxide and followed by a problem
with five pseudo compounds and two reversible reactions. The task of estimating the kinetics
parameter for the reaction is the previous subsection which could also be carried out by
plotting the logarithm of the ethylene oxide concentration as the function of time and then
evaluating the slope of a straight line through the points. This however requires that the
reaction order of ethylene oxide must be assumed before plotting.

When the problems get more complicated for example, for multiple and reversible
reaction, for unknown reaction order and where influence of temperature must be considered,
the determination of rate expression is usually not possible just by shortcut. As an example
considered a system with five pseudo compounds (A, B,C,D and E) in aqueous solution. Two
reversible reactions occurs for the system

Reaction 1: 2A + B C+D

Reaction 2: B + C E

DISTILLATION DESIGN

This section illustrates the use of separation design tool with special attention towards the
equilibrium based algorithm of distillation columns. The mixture will separated originates from
the production of methyl-acetate. Therefore extra distillation column are needed to obtain high
purity product.

9
The composition of the reactor effluent is given. Distillation was choosing as the separation
task for the design details to come to play. When the design algorithm is successful the obvious
next step is to apply the design parameters and the initial estimates in simulations and
distillation column. This is particularly easy in computer aided flow sheeting framework since
the design tool is part of the computer aided system.

SEPARATION SYNTHESIS

The first level calculates differences in pure component properties for each binary pair that can
be combined with the compound in the mixture (called binary property ratio). For each of the
pairs the algorithm identifies the largest property ratio, thereby the most visible separation
technique for the pair is identified. After a screening step, the largest ratio ( which remains)
After screening is identified, this indicates the first separation task to be performed and the
corresponding separation techniques for the flow sheet.

In the second level both pure compounds and mixture properties are
considered in order to simultaneously sequence and select the separation tasks and techniques.
By including mixture properties the second level is able to consider more easily, This
technique requires mass separating agent to give improved estimates for condition of operation.
At the end of the second level a physically feasible separation flow sheet is produced together
with feasible alternatives for each separation task. This feature is especially useful in
combination with structural optimization techniques.

10
 CHAPTER FIVE

CONCLUSION

The development of mathematical models from scratch still error-prone, costly and time-
consuming despite the progress made by commercial modeling tools. The major
accomplishment in commercial tools are improved mechanisms of model structuring and the
provision of easy-to-use graphical user interface. The workflow of modeling is not addressed.
Hence, guidance on request in order to support less experienced modelers is not available. In
order to overcome these problems, a process-centered model development framework has to be
adopted.

11
REFERENCES

1. MODKIT project (Domges et al., 1996)

2. Baumeister .M. (2000) PhD Thesis, RWTH rachen university of Technology.
3. Aspen Technology inc.(2000) ASPEN plus users Guide.
4. Andrecovich , M. and Westerberg .A. (1955). An MILP Formulation For Heat Integrated
Distillation Sequence Synthesis.
5. Haper P.M: Gani, Ishikawa .T. and Kolar .P. (1999). Computer Aided Molecular Design
with Combined molecular modeling and group contribution.

12