DESIGN PROCESS MODEL FOR OPTIMIZING DESIGN
OF CONTINUOUS PRODUCTION PROCESSES
Mika Strömman, Ilkka Seilonen, Jukka Peltola and Kari Koskinen
School of Electrical Engineering, Aalto University, Espoo, Finland
Keywords: Multidisciplinary Design, Process Engineering, Optimization.
Abstract: The non-growing market situation in pulp and paper industry has tightened the competition. Cutting the
design costs by integrating design activities is not going to be enough but the design itself has to be
improved. The design of continuous production processes can be enhanced by utilizing optimization
techniques during the design process. The benefit of the optimization techniques in process design depends
on adequate usage of them during the design process. However, this paradigm shift will require changes in
the existing design processes. In this study, the required changes are identified and a new design process
model describing the optimizing design utilization is developed. The model is then assessed through a case
study and an interview study to ensure that the design process can be realized in the conceptual design phase
of a real delivery project.
1 INTRODUCTION
The market situation in pulp and paper industry have
is setting requirements for the design methods. The
design process itself has to be conducted efficiently,
but in the last years the costs has already been cut
off with better project management and concurrent
engineering. One possibility for rationalization lies
in the design itself; traditionally, the design of the
plant is more oriented into structural design and less
to the optimal combination of operational and
structural design. The design problem can be
formulated as a bi-level multi-objective optimization
problem (BLMOO). Mathematical methods for
solving BLMOO problems exists and the method
have been applied in process facility design in
research projects.
However, the utilization of such optimization
methods requires enhancement of the engineering
process so that the required information for
optimization is available on the right time and the
results of optimization can be used in design. A
design process describing optimizing design of
continuous production processes hasn’t been thus far
presented and it is a necessity for adopting BLMOO-
methods in real delivery projects.
This research has been conducted as a part of a
larger research project in which the objective is to
develop a new optimization based method for
designing a process plant. Our part of the research is
to define a model for optimizing design process and
assess the usability of that model. The research
methods of this study include experimental
definition of a business process model, case study
(with the model) and interview study evaluating the
properties of the model.
In the first chapter the related work and state of
the art is discussed. The following chapter presents
the new engineering business process which takes
into account the optimizing method. Next, a case
evaluating the new engineering business process is
presented and the observations based on expert
interviews are discussed.
2 PROCESS DESIGN AND
OPTIMIZATION
2.1 Design of Continuous Production
Systems
A process plant design is a multidisciplinary process
(process design, automation, software etc.)
(Watermeyer, 2002). Traditionally the process plant
design process has been water fall model like linear
process with stages ending to document deliveries.
492
Strömman M., Seilonen I., Peltola J. and Koskinen K..
DESIGN PROCESS MODEL FOR OPTIMIZING DESIGN OF CONTINUOUS PRODUCTION PROCESSES.
DOI: 10.5220/0003647804920501
In Proceedings of 1st International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SDDOM-2011), pages
492-501
ISBN: 978-989-8425-78-2
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
In a delivery project, the deadlines are counted
backwards from the day that the plant should be
operational. The length of work phases are
determined based on time needed for work and
procurement (Cziner, 2006).
The plant engineering process can be divided
into steps e.g. problem analysis, conceptual design,
detailed engineering, and construction (Tuomaala,
2006). More business oriented divisions are also
possible, for example conceptual phase, pre-
feasibility study, feasibility study, investment
decision and implementation (Diesen, 2007).
Although all the phases are equally important for
reaching the goal, the focus in this research is put on
the conceptual design phase, because the
optimization methods researched in this research
project aim to solve problems on conceptual design
level. Other phases of the engineering process are
relevant to our research in that sense that the tools
and methods should be compatible to the proposed
changes.
In the conceptual design phase a very small
amount of information is available and the time and
resources are limited (Seuranen, 2006). Still the
decisions in this phase fix 80% of the total costs of
the project (Douglas, 1988). Decisions in the early
phases of the project are also quality-critical,
because the costs of changes increase tenfold in each
phase (research – process flow – final design –
production) (Bollinger, 1996). In process plant
engineering, the conceptual design phase is led by
process design. All the other engineering disciplines
are more or less in consulting role. For these
reasons, the greatest advantages can be achieved in
early phases of the business process.
Because of the shortened delivery times, the
other engineering disciplines have to begin their
work before the process design is ready. The saving
using concurrent engineering is calculated to be up
to 50% of the calendar time in a delivery project
(Bañares-Alcantara, 2005).
The sub-processes of any process design task are
design task definition, process structure design,
process operation design an design acceptance.
Process structure design and process control design
interact and should therefore be designed
simultaneously (Pajula, 2006). The existing process
design approaches can be divided to heuristic and
engineering experience based methods, optimization
based methods and case-based reasoning methods
(Seuranen, Pajula and Hurme 2001). Case based
reasoning (CBR) has been applied for design of the
pulp process. The main challenge in CBR is the need
of extensive database to provide the required
knowledge (Pajula, 2006). Outsourcing of the design
work is a common practice nowadays. Fathianathan
and Panchal (2009) have proposed a model to
support outsourcing decisions.
2.2 Optimization in Process Design
Current work practices in forest industry process
engineering are almost solely based on engineering
experience. Simulation and optimization is used in
the design of unit processes, but less in the design of
the process as whole. Plant wide simulation enables
the validation of process structure and control
concepts even before selecting suppliers and
therefore it reduces risks (Ylén, et al, 2005) and
gives a deeper understanding of the process
(Pulkkinen, Ihalainen and Ritala, 2003). According
to the interviews, plant wide simulation is more
useful when building a plant with totally new
concepts when the “rules of thumb” are not
available.
For combining the optimization of plant
structure and plant control, there are several options.
Optimization strategy can be sequential, iterative, bi-
level or simultaneous. (Fathy, Reyer, Papalambros
and Ulsoy, 2001).
Bi-level optimization has been under an active
research lately (Dempe, 2002). Still only a few
research is dealing with multi-objective bilevel
problems. Eichfelder (2010) presents an algorithm
for solving bilevel multi-objective problems. The
combination of dynamic simulator model and
dynamic optimization has been researched for
papermaking process (Linnala, et al, 2011).
2.3 Information Systems for Process
Design
The variety of the Computer Aided Engineering
(CAE) tools supporting process systems engineering
(PSE) is enormous. One of the interviewed
engineering enterprises is using over 50 different
engineering tools. A trend, as seen in modern
integrated process engineering tools, is the
transformation from document-centric design to
data-centric design, realized with database
technology (Comos, 2011),(Smart Plant, 2011),
(Bentley, 2011). Major tool vendors have developed,
acquired and integrated engineering tools from other
engineering disciplines under unified product
families. Modern process engineering support
systems combine modeling and information
management features for engineering of many
aspects of plant engineering, e.g. process, piping,
DESIGN PROCESS MODEL FOR OPTIMIZING DESIGN OF CONTINUOUS PRODUCTION PROCESSES
493
electrical and instrumentation, 3D layout, equipment
lists, part data sheets, etc, thus comprising an
integrated plant information model. This also
enables advanced change management, where
modification of an object through one view notifies
users of other views, looking at the same object.
Multi site work flow management is featured for
both engineering and commissioning. Integration to
external CAE tools is possible through export and
import interfaces using standard or proprietary data
formats. An important prerequisite for cost efficient
integrated engineering is the use of common data
models defined in the standards. ISO 15926,
”lifecycle data for process plant” (ISO 15926) is a
standard dedicated to the process industry, widely
accepted by tool vendors. It has a central role in
pursuing information interoperability between
engineering systems and it is used in many plant
information exchange tool initiatives, such as iRing
(iRing 2011) and XMpLant (XMpLant, 2011) and
even as a native data model of a plant modeling tool
(Bentley, 2011).
Plant information models and semantic
technologies have induced much academic research.
For example, POSC Caesar association (POSC
Caesar, 2011) assembles R&D around the ISO
15926 and modeling methods, such as (Batres, et al,
2007). However, Wiesner, Morbach and Marquardt.
(2010) questions whether a single global plant
information standard is a realistic goal in the first
place and suggest a semantic integration framework
OntoCAPE.
3 MODEL FOR OPTIMIZING
DESIGN PROCESS
Optimizing process design is here modelled in terms
of a business process model. The model describes
the stakeholders of the optimizing process design
and their activities together with the data, knowledge
and utilized mathematical models. Based on these
the requirements for IT support are identified.
Process design as an optimization problem
The process design task can be considered as an
optimization problem. There are a few general
requirements for the process. The process must be
operable, reliable and yield products of sufficient
quality with minimum operational cost. On the other
hand the investment and maintenance cost of the
process should be minimized as well. On this basis it
is natural to consider and model the design problem
as a bi-level multi-objective optimization problem.
The mathematical representation of the general bi-
level multi-objective optimization problem is:
(1)
where
F(x) are the upper level objective functions,
f(x) the lower level objective functions,
G(x), g(x), H(x) and h(x) the upper and lower level
inequality and equality constraints. (Dep and Sinha,
2008)
There are multiple methods for solving bi-level
multi-objective optimization (BLMOO) problems
(Eichfelder, 2010) and (Branke, Dep, Miettinen,
Slowinski, 2008) and the solution method should be
chosen according to the problem itself and the
possibilities for interaction with the decision maker
(Miettinen, 1999). In the plant design process, there
is a logical division to optimization levels, so that
plant structure is the upper level (F(x)) and the
operation of the plant is the lower level (G(x)). The
nature of the plant design is also multi-objetive; the
balancing between design parameters as for example
the total cost of the plant, operational costs,
production quality, production volume and expected
oee-value is difficult and the decision of these values
belongs to the plant owner, not the designer.
Therefore the gathered requirements should also
cover business oriented user preferences.
In this research the solution of the optimization
problem was simplified by scalarizing the lower
level optimization problem, but this simplification
has no affect to this part of the research focusing on
the business process of the design.
3.1 Stakeholders
In the model of optimizing design, new stakeholders,
an optimizer and a modeler, are added to the group
of stakeholders involved in process design as
illustrated in Figure 1. The optimizer is an expert of
mathematical optimization whose responsibility is to
help the process designer in finding more optimal
process designs. The optimizer also needs to
cooperate with the modeler in order to be able to
take into account the operational aspects of the
designed process. These cooperation connections
with the optimizer will also change the work of the
process designer and modeler. Successful
{}
)
(, 1
() 1 ())
() ( )
min () ( (),... ()),
arg min ( ),..., ( ) 0, ( ) 0 ,
() 0, () 0,
,1,...,.
ui
i
xx M
ix mx
LU
iii
Fx Fx F x
subject to x f x f g x h x
Gx Hx
xxxi n
=
∈≥=
≥=
≤≤ =
SIMULTECH 2011 - 1st International Conference on Simulation and Modeling Methodologies, Technologies and
Applications
494
cooperation between the stakeholders is a necessity
for useful design optimization.
Figure 1: Stakeholders of optimizing design.
The role of the optimizer can be described as an
analyst 0. His responsibility is not to make decisions
about process designs but to produce useful
information for the designer about possibly better
designs. In order to do this, the optimizer will need
to have expertise in multi-objective optimization and
familiarity in process design. The adoption of
optimization also changes the roles of preexisting
stakeholders. Designer is the decision-maker of the
process design and the client of the optimizer. In
optimizing design the designer has to select a part of
his design problem for optimization together with
the optimizer. In addition to this, the designer also
has to cooperate with the optimizer during the
optimization process and finally interpret the results
and decide how to apply them. Again, the optimizer
will become the client of the modeler.
In order to adopt the new business process, all
the stakeholders should gain some advantage of the
enhanced business process. The process designer
gains competitive advantage by offering design that
is more tailored and more cost effective along the
life cycle of the plant. For the optimizer and
modeler, the new model opens a totally new
business possibility.
3.2 Business Processes
Dynamic and stochastic nature. In this subtask the
designer and the optimizer can rely on the expertise
In the model of optimizing design the activities of
process design have partially changed. The basis for
the activities is the existing design processes that are
extended and partly modified. The suitable time for
optimization is the conceptual design phase. When
the designer identifies a need for optimization in his
conceptual design, he initiates cooperation with the
optimizer. During this cooperation an optimal design
balancing both structural and operational aspects of
the design are being searched for. This process can
be described as expert cooperation in which also the
modeler will be included.
Figure 2: Activities of optimizing design.
The optimization activities take place in a few
stages as an extension to conceptual process design
phase as illustrated in Figure 2. The process starts
from optimization problem definition and continues
through optimization problem-solving until result
interpretation. During these stages different
cooperation patterns between the designer, optimizer
and modeler are needed. The whole process and
each of its stages may also be iterative.
The purpose of the optimization problem
definition is to define a part of the designer's design
problem as BLMOO for the optimizer. This stage is
performed by the designer and optimizer together.
The designer identifies parts of the overall design
problem in which balancing structural and
operational aspects of the design is essential. The
solvability of the problem is then assessed by the
optimizer, designer and modeler together. The
assessment requires expertise of all three
stakeholders because the result depends not only on
the problem itself but e.g. optimization tools,
process models and data about the process.
Eventually the designer and the optimizer should
agree on a useful and solvable design optimization
problem, which the optimizer then formulates as a
BLMOO problem.
An important subtask of the optimization
problem definition is process operation modeling.
Modeling the operational part of the design problem
is much more difficult than the structural part due to
its of the modeler. The modeler is expected to have
expertise about both mathematical modeling and the
DESIGN PROCESS MODEL FOR OPTIMIZING DESIGN OF CONTINUOUS PRODUCTION PROCESSES
495
designed process itself, i.e. its chemical and physical
characteristics. Based on his expertise the modeler
should be able to create such operational models that
are suitable to be used in optimization. The
suitability of the models will be assessed by the
optimizer and the designer.
The stage of problem-solving is focused on the
optimizer. However, cooperation with the other
stakeholders is likely to be needed also in this stage.
In the beginning of this stage the data and models
required in the optimization are expected to be
transferred to the optimizer in a form which he can
utilize. Depending on the utilized MOO method,
different type and amount of cooperation with
designer will be needed also during the actual
problem-solving. According to an interview (see
chapter “Interviews”) industrial experts seem to
favor optimization methods which lead to
representations of Pareto optimal designs.
The last stage of optimizing design is result
interpretation. Also this stage is performed in
cooperation between the designer and the optimizer.
The optimizer prepares result presentations, which
indicate Pareto optimal designs and help the
designer evaluate the impact of his preferences on
the design. The designer is expected to study the
design optimization result, assess its reliability and
make decision about possible changes to his design.
This is not necessary a straightforward task and is
likely to require assistance from the optimizer and
the modeler. The reliability of the optimization
result is dependent on used operational models and
data. Sensitivity analysis of the result might also be
needed. In the end, the designer can adopt changes
to his design or reject the optimization results and
reformulate the optimization problem with the
optimizer.
3.3 Data, Knowledge and Models
The optimizing design requires additional
knowledge, data and models than the state-of-the-art
approaches to process design. The new requirements
originate from the need to solve the process design
BLMOO problem. The new requirements for
knowledge, data and models in optimizing design
are summarized in Table 1. In addition to these, the
previous requirements are still valid, e.g. designer
knowledge for process design, use of design data
and design models.
The expertise and knowledge of the stakeholders
involved in optimizing design is complementary.
The designer has knowledge about industrial
processes and their design, customer requirements
and evaluation of process designs. Meanwhile, the
modeler is expected have knowledge about similar
processes and their mathematical modeling.
Table 1: Knowledge, model and data requirements in
optimizing design.
Knowledge Data Models
Designer
Process design,
process
knowledge, some
understanding
about
optimization
Design data,
customer
requirements
Flow diagram
P&ID
Plant Model
Optimizer
Optimization,
some
understanding
about design
Design data
and operational
data from
designer and
modeler
Operational
and design
level problem
formulation
models for
optimization
Modeler
Modeling,
Process
knowledge
Operational
data, some
design data
Operational
models (e.g
break
probability
model,
The knowledge of the optimizer concerns about
optimization and acting as an analyst in a decision-
making process of MOO. However, during the
activities of the optimizing design combination of
the knowledge of different stakeholders and
knowledge transfer between them is necessary. A
partially common understanding of the design
problem shared by the stakeholders has to be created
(Konda, Monarch, Sargent and Subrahmanian,
1992). This is may be done according to the
BLMOO of the process design.
Mathematical models of the designed process
have an important role in optimizing design. Models
are needed particularly for modeling the operation of
the process. Mathematical models have been used in
the design of continuous processes also previously,
e.g. in simulations (Ylén, et al, 2005), but these
models are not necessary suitable to be used in
optimizing design. In order to be able to be utilized
in optimizing design, the operational models need to
have a suitable balance of modeling capability and
computational requirements. The computational
requirements can be met by modeling only selected
parts of the process. More precise models may be
utilized after the design in a design validation stage.
The optimizing design requires data transfer
between the optimizer and other stakeholders, which
is not needed without optimization. The most
important data transfer takes place from the designer
to the optimizer. The designer has to pass the most
SIMULTECH 2011 - 1st International Conference on Simulation and Modeling Methodologies, Technologies and
Applications
496
of the data describing the design optimization
problem to the optimizer, e.g. flow diagrams,
dimensions of equipments etc. The other source of
data to the optimizer is the modeler. He is expected
to deliver to the optimizer the operational models
and the data required by them, e.g. model describing
the probability of break. This data is intended for
algorithmic processing, which indicates a
requirement for adequate precision. The final data
transfer consists of the optimization results, which
are passed from the optimizer to the designer. This
data has a form of a document. A major requirement
for it is understandability.
3.4 Requirements for Information
Systems
The new requirements for the information systems
mainly rise from the new data flows between
designer, optimizer and modeler. The amount of data
from the designer’s plant model can be quite huge,
so the optimizer should have access to the designers
plant model tool to be able to import the needed set
of design data. A new thing is that the designer
should also include the constraints of the design to
the model when applicable. The design data should
be transferable to optimizing tool as well as the
models that the modeler has created. The support for
representing the alternatives to the designer is not
that critical, because that document should be kept
brief and simple.
4 ASSESSMENT OF THE MODEL
In this chapter, the business process model of
optimizing design is assessed through a small-scale
case study. This case study was carried out as a part
of a wider research project and the results of the
mathematical solution of the BLMOO in this case
can be found in our partners’ publications
(Ropponen et al., 2010), (Eskelinen, et al., 2010),
(Ropponen et al., 2011) and (Ropponen, Rajala,
Ritala, 2011). The case was evaluated by internal
review and expert interviews.
4.1 Case Study
4.1.1 Case Design Problem
The design task in the case study was to dimension
six storage towers of a part of a paper-making
process and to guarantee the runnability and stability
of the process. The dimensioned storage towers
Figure 3: Flow diagram of the process in case study.
include TMP (thermo-mechanical pulp), chemical
pulp, wet broke, dry broke, clean water and 0-water.
The design problem is illustrated in Figure 3 and
further explained in (Ropponen, Rajala, Ritala
2011).
4.1.2 Stakeholders
The actors involved in the design process in the case
study include the designer, optimizer and modeler.
The roles were manned by research teams involved
in the project.
The designer had the main responsibility of the
project. He carried out the requirement elicitation
with the end-customer, proposed a conceptual design
and initiated the problem formulation for the
optimization. He then had a key role in data
acquisition for the model building. After getting the
optimization results, he made the decisions
according to the end-users preferences.
The optimizer participated in the problem
formulation by having an opinion what kind of
problems can be solved with optimization. After the
problem formulation, the optimizer then asks the
modeler to build necessary models for optimization
and then chose the right optimization method.
Finally, a suitable method for presenting the results
was chosen.
The modeler was responsible for creating a
model simple enough to be calculated. The modeler
was also responsible to make sure that the
simplifications do not affect to the problem to be
solved.
TMP Chem pulp
Broke
(deficit/
surplus)
Clean water
Separation
0-water
Paper Machine
Quality
control
Dry broke
storage
DIL DIL
DIL DIL
DIL
DIL
DIL
Mixing
TMP
Pulpers
Fresh
water
Fresh
water
Evaporated
water
Paper,
net
Paper,
brutto
Filler
DESIGN PROCESS MODEL FOR OPTIMIZING DESIGN OF CONTINUOUS PRODUCTION PROCESSES
497
4.1.3 Business Process
The project could be divided into four main tasks:
problem formulation, model building, problem
solving (optimization) and result interpretation.
Problem Formulation
At the starting point of the case study a part of the
conceptual design was already performed, e.g. the
number of storage towers and material flows
between them was defined. The designer and
optimizer then discussed the possibilities for a
manageable optimization problem. They designed
that the optimization activity concerns only about
operation design and the dimensioning part of the
structure design. Also the amount of optimized
parameters was reduced in negotiations between the
designer and the optimizer. During the optimization
activity a mathematical model of the problem was
created and used for finding an optimal design under
the specified requirements. The design problem was
formulated as follow:
{}
{}
{}
{}
{}
4
,max)
1
2
0,
2
0,
2
2
()
(() )
(() )
min
(())
(( 1) ())
i
i
Filler Filler
bw bw
d
strength
v
HV
Eq nq
Eqnq
Eq n
Eun un
ET
ψ
ψ
ψ
ψ
ψ
=
⎧⎫
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎨⎬
⎪⎪
⎪⎪
⎪⎪
+−
⎪⎪
⎪⎪
⎩⎭
(2)
where (
,
) is the investment cost of the 4
selected tower volumes,
is the time till one of the
towers goes empty or flows over, and E
Ψ
{} denotes
the expectation value of the system performance as
Ψ is the stochastic process with applied dosage
policy. (Ropponen, Rajala, Ritala, 2011)
The operational problem, i.e. the lower level of
the BLMOO was formulated as:
2
0,
1
2
0,
1
2
0,
1
2
1
,max
()
1..
,min
(
1..
()( ( ) )
()( ( ) )
()( ( ) )
min
()(( ) ( 1)
(( ) )
max
()
(( ) )
max
H
H
H
H
H
H
K
Filler Filler
k
K
bw
k
K
strength strength
k
K
u
k
ii
up
kK
i
ii
do
kK
i
kq n k q
k qbw n k q
kq n k q
kun k un k
pV n k V
pk
pV n k V
p
γ
γ
γ
γ
=
=
=
=
=
=
+−
+−
+−
+− +
+>
+>
)
,min ,max
()
.
wn
jjj
k
st u u u
⎧⎫
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎨⎬
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎪⎪
⎩⎭
≤≤
(3)
where

,

, and

are the quality
variables with
s as their nominal values,
is the
optimization horizon,
γ
(k) a time-wise weighting
factor, u is a vector of pulp/water flows to be
controlled,
()
(
)
and
()
(
)
are the accepted
risks for a tower overflow/goes empty k time steps
from the present time n defined as
(/)
(
)
=
1−1−
(/)
, i refereeing to the storage towers
for clean water, 0-water, broke, and dry broke. V
i,max
is the volume of the ith storage tower, i.e. the
maximum amount of pulp/water in the tower, and
V
i,min
is the minimum amount of pulp/water in the
tower. U is the control variable describing the broke
dosage from the broke tower to the system
(Ropponen, Rajala, Ritala, 2011).
Simplified, on the operational level we optimize
the variances of the quality attributes of the paper
and the broke dosage and the probability of
under/overflows. On the design level, we optimize
the design according to the sizes of the tanks and
expectation values of the system performance.
Model Building
At the same time that the optimizer negotiated with
the designer about the problem formulation, he had
to discuss with the model builder if a suitable model
for the problem can be built. In this discussion there
were two main themes: is the physical phenomenon
of the problem known or is there enough data to
model the problem stochastically and can the model
be simple enough that it can be calculated fast
enough in the optimization loop.
Optimization and Result Interpretation
In this case example, the tasks of problem
formulation, model building and optimization were
performed simultaneously and were highly iterative.
The main focus of the case example was in
optimization. The results of the optimization are
described in (Ropponen et al., 2011) and (Ropponen,
Rajala, Ritala, 2011).
After the optimization, the results were presented
to the designer as two-dimensional Pareto optimal
sets. In Fig.3, a Pareto optimal set in respect to the
two most important parameters is presented. The
designer then made the decisions e.g. between a
decent investment cost and an acceptable probability
of break.
4.1.4 Data, Knowledge and Models
The designer in this case had a wide experience in
process design, paper making, modeling and
optimization. The optimizer was mathematically
oriented, but had only minor experience on paper
SIMULTECH 2011 - 1st International Conference on Simulation and Modeling Methodologies, Technologies and
Applications
498
making or process design. The modeler was familiar
with process modeling and optimization.
The largest data flow in the process was from
designer to optimizer. The designer had to
communicate the customer requirements, the
original design about the structure and operation and
the freedoms and limitations for optimization in the
design. The main models for this communication
were a process flow sheet and steady-state model of
the process. Making of these models was mainly a
task for the designer. The designer was able to
formulate most of the limitations and requirements
in numerical form, e.g. the probability of the break
may not be greater than Pmax. Due to the nature of a
first time project, the data transfer between the
optimizer and modeler was also huge.
Modeller was responsible for building three
models: dynamic model, predictive model and a
validation model. The two first mentioned were used
in optimization while the validation model build
with different simulation software was used only for
one selected design.
Practically, the problem formulation and
optimization required simultaneous model
development, because there wasn’t previous
knowledge about feasible models.
The results of the optimization were delivered as
a document containing simulation graphs and Pareto
optimal sets (one example in Figure 4) of
optimization results.
Figure 4: Design solutions in respect to investment cost
and time until production stop. Pareto optimal set of
designs circled. (Ropponen, Rajala and Ritala, 2011).
4.1.5 Information Tools
This case example was carried out as a research
project, and therefore the engineering tools used
didn’t match the ones used in industry. MATLAB
was used both for the optimization and simulation
for optimization. APROS process simulator was
used in validating the results of optimization.
4.2 Interviews
In order to get information of the process
engineering business process today and to validate
the proposed changed to the process in order to
adopt a new optimizing design process, a set of
interviews were performed. The interviewees
represented actors in both chemical and pulp &
paper industries and contained process designers,
automation designers and IT-system experts in
process design companies and engineering
enterprises. In addition a simulation expert and an
optimization expert were interviewed.
The topics of the interviews were motivation and
feasibility of optimizing design, current design
practices vs. optimizing design and IT systems vs.
requirements of optimizing design.
The following observations could be made about
issues concerning the motivation and feasibility of
optimizing process design:
There are business requirements to decrease the
costs of plant design projects. At the same time the
quality of the design should be increased and cost
decreased. The effect of optimizing process design
process on all three aspects (design quality, design
cost, project cost) should be taken into account.
The process design practices in different
industries are heterogeneous. In paper and pulp
industry process design can be characterized as
engineering-oriented, i.e. an engineering design
system is the primary design tool. As a comparison,
in chemical industry process design is quite
simulation-oriented, i.e. a simulator is the primary
design tool. The design practices of chemical
industry are closer to the optimizing process design
process than the ones in paper and pulp industry.
The following observations could be made about
issues concerning the differences between current
design practices and optimizing design:
Cooperation between different parties involved
in a design project has recently been emphasized by
engineering companies. Cooperation is needed for
the efficiency of a design process, e.g. finding out
the requirements of the customer early enough,
ensuring consistency of the designs from different
designers and handling the effects of design
changes. The optimizing design process should fit to
the cooperation practices.
The design of a process is divided to several
designers according to different systems or parts of
0
2
4
6
8
10
12
7000 8000 9000 10000 11000 12000
1/time (x10E-3)
Investment cost
DESIGN PROCESS MODEL FOR OPTIMIZING DESIGN OF CONTINUOUS PRODUCTION PROCESSES
499
the process. This is done due to the different
expertise of the designers and concurrency of the
design work. There are usually some buffers in the
design between the designs by separate designers.
From the optimization viewpoint this division is
questionable. The optimizing design process is likely
to change the division of work.
The division of work is also reflected to current
optimization practices. They are optimizing unit
processes rather than the whole process. The
optimizing design process should change this
practice, too.
The trust of the customer on the feasibility of the
process design in a very important issue, which is
affected by many factors, e.g. references of the
vendor and difference of the design to existing ones.
It was mentioned that particularly in the paper and
pulp industry customers do not trust simulations as a
process design validation tool. Validation of the
design results should be a primary concern also in
optimizing process design.
The following observations could be made about
issues concerning the differences between
requirements for current IT systems and IT systems
when using optimizing design:
The IT-architecture of an engineering company
is usually quite heterogeneous, i.e. there are several
different IT-systems used during a design project.
Sometimes there are even several alternative IT-
systems for same design tasks, e.g. due to customer
requests. The heterogeneity of IT-systems may
hinder the implementation of IT-support optimizing
design.
There is a slowly progressing shift from
document-centered design paradigm to data-centered
design paradigm in plant design. The optimizing
process design process should be made to fit the data
model -oriented design paradigm because its
meaning seems to be increasing in the future. It is
also likely to be more suitable basis for optimizing
design than the older document-oriented design.
5 CONCLUSIONS
In this paper a business process model for
optimizing design in continuous process facility
engineering has been presented. This model was
considered from the viewpoints of stakeholders,
process, knowledge, data, models and tools. The
model for optimizing design was then assessed by
applying it in an experimental case study and by
interviewing experts.
Based on this study, a few conclusions can be
made. The greatest change is the new roles of
optimizer and modeler, which make the process
more iterative between optimizer and process
designer. The new roles require a shared knowledge,
because the work can be described as expert co-
operation. The business process in optimizing design
is more iterative than in traditional design because of
the need for negotiation in the problem formulation
and the uncertainties in the modeling. In addition to
this, the interviews also illustrate the importance of
validation of process designs. Validation of the
designs so that the customer will trust them is a
primary concern to be observed in future research.
It must be noted that the design business process
in this paper is presented at a general level and it
must be specified when used as actual process. In
future, the process model is evaluated and specified
in a larger case study.
ACKNOWLEDGEMENTS
This research was supported by Forestcluster Ltd
and its Effnet program.
REFERENCES
Bañares-Alcantara, R., Concurrent process engineering,
state of the art & outstanding research issues,
http://cape-alliance.ucl.org.uk/CAPE_Applications_
etc/Initiatives_and_Networks/About_CAPENET/Key_
Research_Areas/CPE_Main_Directory/CPE_State_of_
the_Art.pdf 2005
Batres R., West M., Leal D., Price D., Masaki K., Shimada
Y., Fuchino T., Naka Y., An upper ontology based on
ISO 15926, Computers and Chemical Engineering 31
(2007) 519–534
Belton V., Steward T., Multiple criteria decision analysis –
an integrated approach. Kluwer Academic Publishers
2001
Bentley, http://www.bentley.com/
Bollinger, R. E. Clark, D. G., Dowell, A. M. III, Ewbank,
R. M., Hendershot, D. C., Lutz, W. K., Meszaros, S. I.,
Park, D. E., Wixom, E. D., Inherently Safer Chemical
Processes, A Life Cycle Approach, Center for
Chemical Process Safety/AIChE, 1996
Branke J. Deb K., Miettinen K., Slowinski R.,
Multiobjective Optimization Interactive and
Evolutionary Approaches
Comos, http://www.comos.com/
Cziner K. Multicriteria Process Development and Design.
Doctoral dissertation. Helsinki University of Techno-
logy 2006. ISBN 951-22-8429-4
SIMULTECH 2011 - 1st International Conference on Simulation and Modeling Methodologies, Technologies and
Applications
500
Dempe S. Foundations of Bilevel Programming. Noncon-
vex Optimization and its Applications, vol. 61.
Kluwer, Dordrech, 2002.
Dep K., Sinha A., Solving Bilevel Multi-Objective
Optimization Problems Using Evolutionary Algori-
thms http://www.iitk.ac.in/kangal/papers/k2008005.
pdf
Diesen M. Papermaking science and technology. 1, Eco-
nomics of the pulp and paper industry. Finnish Paper
Engineers’ Association, 2007.
Douglas, J. M., Conceptual Design of Chemical Process,
McGraw-Hill, New York 1988
Eichfelder G., Multiobjective bilevel optimization,
Mathematical Programming 123, 419-449, 2010
Eskelinen P., Ruuska S., Miettinen K., Wiecek M.,
Mustajoki J. A Scenario-Based Interactive Multiobje-
ctive Optimization Method for Decision Making under
Uncertainty. 25
th
Mini-EURO Conference,
Uncertainty and Robustness in Planning and Decision
Making Under Uncertainty. Coimbra, Portugal 15-17
April 2010.
Fathianathan M., Panchal J., Incorporating design outsour-
cing decisions within the design of collaborative
design process. Computers in Industry 60 (2009)
Fathy H., Reyer J., Papalambros P. and Ulsoy A., On the
Coupling between the Plant and Controller
Optimization Problems. Proc of the American Control
Conference, Arlington, VA, June 25-27 2001.
iRing User Group, http://iringug.org/
ISO 15926, Industrial automation systems and integration
– Integration of life-cycle data for process plants
including oil and gas production facilities,
http://15926.org/
Konda, S., Monarch, I., Sargent, P., & Subrahmanian, E.
(1992). Shared memory in design: a unifying theme
for research and practice. Research in Engineering
Design, 4, 23–42.
Linnala M., Ruotsalainen H., Madetoja E., Savolainen J.,
Hämäläinen J. Dynamic simulation and optimization
of an SC papermaking line – retention disturbance as a
case study. Nordic Pulp and Paper Research Journal.
Sent for review 2010.
Miettinen K., Nonlinear multiobjective optimization,
Kluwer Academic Publishers, Boston 1999
Pajula E., Studies on computer aided process and equip-
ment design in process industry. Doctoral dissertation.
Helsinki University of Technology 2006. ISBN 951-
22-8489-8. Available: http://lib.tkk.fi/Diss/2006/isbn9
512284898/
POSC Caesar Association, http://posccaesar.org/
Pulkkinen P., Ihalainen H., Ritala R., Developing Simula-
tion Models for Dynamic Optimization, SIMS 2003 –
44th Conference on Simulation and Modeling on
September 18-19, 2003 in Vesterås Sweden. available:
http://www.scansims.org/sims2003/papaers/SIMS200
3_12.pdf
Seuranen, T. Pajula E., Hurme M., Applying CBR and
object database techniques in chemical process design,
Lecture Notes in Artificial Intelligence, VOL 2080,
731-743, 2001.
Seuranen T. Studies on Computer-Aided Conceptual
Process Design. Doctoral dissertation. Helsinki
University of Technology 2006. ISBN 951-22-8267-4.
Available. http://lib.tkk.fi/Diss/2006/isbn9512282674/
Ropponen A., Rajala M., Ritala R. Multiobjective
optimization of the pulp/water storage towers in
design of paper production systems. 21
st
European
Symposium on Computer Aided Process Engineering
– ESCAPE 21, Chalkidiki, Greece 2011(In Press).
Ropponen A., Ritala R., Pistikopoulos E. N. Broke mana-
gement optimization in design of paper production
systems. ESCAPE 20, Ischia, Naples, Italy, June 6-9,
2010.
Ropponen A., Ritala R., Pistikopoulos E., Optimization
issues of the broke management system in
papermaking. Computers and Chemical Engineering
(2011).
Smart Plant, http://www.intergraph.com/
Tuomaala M. Conceptual Approach to Process Integration
Efficiency. Doctoral dissertation. Helsinki University
of Technology 2006. ISBN 978-951-22-8788-8.
Available: http://lib.tkk.fi/Diss/2007/isbn9789512287
888/
Watermeyer P. Handbook of Process Plant Project
Engineers. Wiley 2002. ISBN 1-86058-370-9
Wiesner A., Morbach J., Marquardt W., Information inte-
gration in chemical process engineering based on
semantic technologies, Computers & Chemical
Engineering,
doi:10.1016/j.compchemeng.2010.12.003
XMpLant, http://groups.google.com/group/xmplant
Ylén J., Paljakka M., Karhela T., Savolainen J, Juslin K.,
Experiences on utilising plant scale dynamic simula-
tion in process industry. Proc of 19
th
European confe-
rence on modeling and simulation EMCS 2005. http:
//www.copm.glam.ac.uk/ASMTA2005/Proc/pdf/sim-
serv-03.pdf
DESIGN PROCESS MODEL FOR OPTIMIZING DESIGN OF CONTINUOUS PRODUCTION PROCESSES
501