Escolar Documentos
Profissional Documentos
Cultura Documentos
G. Schembecker et al.
Tim Seifert
Anna-Katharina Lesniak
Research Article
Stefan Sievers
Gerhard Schembecker
Christian Bramsiepe
In the light of market uncertainties and globalization, the capacity of fast and
flexible adaption of a chemical plant to changing market conditions becomes a
major necessity in future production scenarios. The capacity flexibility of a plant
can be influenced by oversizing and numbering up of unit operations. The additional costs for these measures have to be compensated for by the financial benefit
arising from the option to adapt to uncertain market demands. Therefore, the additional costs have to be balanced against the gain in flexibility. In this work,
methods for the evaluation of capacity flexibility are adapted from the manufacturing industries to enable a cost-effective design of a modular plant. The evaluation methods are used to select modular plant setups allowing the best compromise between costs and flexibility in different market scenarios.
Keywords: Capacity, Expansion, Flexibility, Modularization, Plant design
Received: September 16, 2013; revised: November 20, 2013; accepted: November 26, 2013
DOI: 10.1002/ceat.201300635
Introduction
www.cet-journal.com
Plant design
333
sary as each unit operation has its own operating range and
A plant with high volume flexibility allows the adaption of
impact on the costs structure. Considering these two points
the plant capacity to market changes without changing the
allows a better control of the size and cost of each expansion
setup, e.g., by adding additional units [5]. Thus, the operating
step.
window of the plant set by the minimal and the maximal reaThe aim of this work is to present a method capable of idenlizable capacity is a measure of volume flexibility. To allow a
tifying the best compromise between capacity flexibility and
comparison between different designs, Gerwin [5], Cheng et
investment costs in a modular plant design. This method
al. [6] and Parker and Wirth [7] developed methods to evalushould be applied in the early stages of process design when
ate the volume flexibility of manufacturing systems. In chemthe decision for or against a modular plant has to be made.
ical engineering, volume flexibility can be related to operaTherefore, methods allowing a quick estimation of the capacity
tional flexibility. This type of flexibility describes a parameter
flexibility with low computational effort are needed. The evalrange under which a plant can be operated by maintaining
uation methods are adapted from strategies in the manufacturspecifications like product purity [8, 9]. In this context,
ing industries. These evaluation methods will be used in a case
Grossman et al. [10] developed a measure for the feasible
study of the continuous production of a recombinant protein.
operating range of a plant for a set of uncertain process
Lastly, it will be shown which method can be applied in differparameters like kinetics or heat transfer coefficients. Based on
ent market scenarios.
this idea, multiple variations of algorithms have been used to
optimize operational flexibility [1113]. Lima et al. [14]
investigated the application of operational flexibility analysis
2 Case Study
for a wide range of example problems for steady-state and
dynamic systems. Malcolm et al. [15] showed the value of
Here, the model process, i.e., a plant for the production of a
combining process design and design of plant control to optirecombinant protein, and different market scenarios for the
mize the operational flexibility of a plant. These publications
following evaluation will be described. A detailed process
allow designing a process that covers a wide uncertainty range
description, the required simulation work, and the assumpof different process parameters under optimal costs. The
tions made can be found in [23]. In the following, only fundamethods can also be used to optimize the operability of the
mental information will be presented.
plant at different capacities. Due to the high complexity of
the methods, the implementation effort for larger process
design is very high.
2.1 Example Process
Expansion flexibility describes the opportunity to adapt the
production output if the market demand increases durably
The flow chart of the process can be seen in Fig. 1. The protein
[3]. In contrast to volume flexibility, additional apparatuses or
is produced by Saccharomyces cerevisiae (bakers yeast) with a
production lines are installed to follow the market demand. In
capacity of 10 t a1 in continuous fermentation (FER). To reach
the field of goods manufacturing industries, evaluation meththe desired cell concentration, the continuous fermentation
ods for expansion flexibility have already been developed
process is usually started in batch mode, followed by a period
[16, 17]. In the chemical engineering literature, the focus lies
of fed-batch operation. A further continuous mode is started
on optimization of expansion strategies based on expected
until the process has to be aborted to avoid gene mutation.
market development and investment costs. The effects of the
The operating costs of continuous fermentation consist of the
operability of the plant at different capacities are often of subcosts for coolant and electric power, oxygen, water, substrate,
ordinate importance. Oldenburg et al. [18] and Wiesner et al.
and cells.
[19] elaborate on the advantage of installing multiple smaller
plants compared to a single plant built for design
capacity. Coleman and York [20] set up a model
for optimizing investment strategies in uncertain
markets. Lier and coworkers showed the benefit of
stepwise plant expansion using discounted cash
flow (DCF) calculations [21] and a real-option
analysis approach [22]. These methods can be used
to find an optimal investment strategy for adapting
to an increasing market with multiple but identical
production lines. The drawback is that economies
of scale limit the adaptability as the investment
costs for multiple smaller plants are much higher
compared to one large-scale plant. Thus, the financial benefit from stepwise expansion is limited by
the high additional costs, and high capacity flexibility cannot be achieved. Here, the modular
approach offers additional options to overcome
this limitation. Therefore, each unit operation has
to be taken into account separately. This is necesFigure 1. Flow sheet of the model process.
www.cet-journal.com
334
G. Schembecker et al.
www.cet-journal.com
Plant design
335
Thus, the basis for the evaluation of volume flexibility must be an investigation of the operating
window of each unit operation.
Cheng et al. [6] described such an evaluation of
operating windows for flanged shafts. In their
example, volume flexibility depends on the number of shafts that can be produced in a certain
time. The production time in one process step for
one shaft depends on the time needed to achieve
the final diameter of the shaft on a turning lathe.
Furthermore, it has to be ensured that the following machines can handle the variety in shaft diameters and have a capacity matching with the other
machines in the manufacturing process. These two
Figure 2. Operating cost structure (left diagram) and investment cost structure
characteristics set the feasible capacity range [6].
(right diagram) optimized for a capacity of 10 t a1.
This idea can be transferred to the field of chemical engineering. However, the number of produced pieces has to be translated into a production output in
mass per time as the measure for the feasible capacity range.
This capacity range is limited by apparatus design and process
requirements. To keep an operating point inside these boundaries, the process parameters have to be adapted. The individual operating windows of each apparatus can be combined to
get the operating envelope of the overall plant as it was presented by Cheng et al. [6]. This operating envelope is set by
the unit operation with the lowest maximum and the highest
minimum capacities. The wideness and the values of the
bounds of the operating envelope depend on the selected apparatus type and the size and number of the parallel apparatuses of each unit operation in the process. Thus, the module
combination is the lever to influence volume flexibility. Fig. 4
Figure 3. Market development for the investigated scenarios.
shows an example for the determination of an operating envelope from operating windows.
ods, which will be presented in the next section, must be to
investigate the tradeoff between the achieved flexibility and the
resulting increase in investment costs.
Flexibility Analysis
MF [m2]
DF [m2]
UF1 [m2]
UF2 [m2]
1 40
1 20
15
8 20
2 20
2 0.36
7.1 million
2 25
2 10
61
10 20
5 10
4 0.36
10.3 million
3 15
2 10
61
10 20
5 10
4 0.36
11.4 million
Setup
To set the change in volume flexibility in relationship to the necessary expansion, a mathematical
description for volume flexibility is
needed. In this way, a tradeoff
between the costs for changes in
plant design by numbering up and
www.cet-journal.com
336
G. Schembecker et al.
Cmax Cmin
;
Cmax
(2)
3.2
Expansion Flexibility
Evaluation of the expansion flexibility EF can be used to identify the plant design that allows the most cost-efficient and
flexible adaption to a long-term market development. In the
following, two evaluation methods for EF will be adapted from
methods used in the manufacturing industry.
The first method was developed by Rogalski [17] who proposes a capacity-based method to rate EFRogalski comparable to
the factor for volume flexibility described above (Eq. (2)).
Rogalski describes the value for EFRogalski as the difference
between the target capacity Ctarget and the new breakeven
www.cet-journal.com
Ctarget CBreakEven;new
;
Cmax;old CBreakEven;old
(3)
Figure 5. Breakeven diagram describing the change in the flexibility range according to Rogalski [17].
PE
PI
PN
PN
(4)
Plant design
rNPV
1 r
(5)
EFJacob
EAAE
EAAI
EAAN
EAAN
337
(6)
As the future market development will differ from the forecasts made during plant design, it is necessary to take different
scenarios into account. These scenarios represent different possible market developments. EF is calculated for different setups
in each of the scenarios. By this, it is possible to evaluate which
of the setups achieves the highest EF in most of the market scenarios. Furthermore, it is possible to take market decreases
into account. In this way, also a stop of the expansion plan can
be respected and evaluated.
Results
4.1
www.cet-journal.com
338
G. Schembecker et al.
The two-stage ATPE is carried out in two horizontal mixer-settler units. For total phase separation a certain settling time is needed. To ensure
this settling time, it must be guaranteed that the
settling velocity of the dispersed droplets is lower
than the velocity of the continuous phase [28].
The needed settling time is provided by the design
of the settler by adapting the length, diameter, and
position of the interphase. As the velocity of the
continuous phase depends on the volume flow of
the feed stream, the residence time must be higher
at higher loads. As the settler design limits the settling time, overdesign and numbering up must be
Figure 8. Resulting operating envelope for each alternative.
used to operate the settler at higher loads as well.
The capacity of the mixer is limited by a minimum
residence time that has to be achieved for good mixing [28].
leads to lower investment costs. Thus, the difference in investIn case of apparatuses operated in parallel, it is possible to
ment costs between setup 1 and the other setups is mainly gentemporarily shut down some of the units. This opportunity
erated by economies of scale with regard to the fermenter.
has to be taken into account to estimate the lower bound of
However, using only one fermenter limits the lower bound of
the operating window. Fig. 7 shows the achievable capacity
the operating envelope of setup 1 compared to the other setlimits for continuous fermenters operated in parallel, with difups. Furthermore, oversizing in setup 1 is low so that the upper
ferent sizes and numbers of parallel units. To keep the resibound of the operating envelope is also smaller compared to
dence time constant, the filling level can be adapted in the
the other setups. Setup 3 has the widest operating envelope,
range between 50 and 85 % of the total volume. Compared to
but at the same time the highest investment costs of the setups
one 40-m3 fermenter, it is possible to reduce the lower reaction
investigated.
volume from 20 to 4 m3 if five 8-m3 fermenters are used. Using
For the example process, the breakeven capacity is approxitwo 25-m3 fermenters enables an increase of the maximal reacmately 4.6 t a1 for setup 1 and 5.0 t a1 for setups 2 and 3. Thus,
tion volume to 42.5 m3. From this example, it can be seen that
the operating envelopes of setup 2 (Cmin = 4.8 t a1) and setup 3
the apparatus size and numbering up are the degrees of free(Cmin = 3.8 t a1) are limited by their breakeven capacity. Setdom to influence the volume flexibility.
up 1 has a minimum capacity Cmin of 7.7 kt a1. Consequently,
the volume flexibility generated by numbering up cannot be
fully used and additional investment for higher flexibility does
not totally pay off.
Setup 1 has a VF of 0.4, which is the lowest of all setups.
With an additional investment of $ 3 million for setup 2, a VF
of 0.68 can be achieved, which is a significant increase of the
operating envelope. Setup 3 has a VF of 0.65, which is lower
than the one of setup 2. The reason for this is the limitation of
the lower bound of the operating envelope by the breakeven
capacity. Therefore, designs with a Cmin below the breakeven
point should be avoided to allow for an economic utilization
of the installed equipment also at the lower bound of the operating envelope.
Based on these results, it is possible to select a setup dependent on the uncertainty in market demand. The operating en3
Figure 7. Fermenter capacity for different sizes (8, 25, and 8 m )
velope of the plant must be high enough to cover the range of
and the number of parallel units (1, 2, and 5) depending on the
uncertainty. In cases with a low market uncertainty, setup 1
filling level.
should be preferred as its investment costs are the lowest of all
setups. If high uncertainties in market demand are expected,
Fig. 8 shows the resulting operating envelopes for each setup
setup 2 should be selected as it provides a compromise between
presented in Tab. 3. The three radar charts show the differences
the flexibility and costs of the three setups. Setup 3 has the
between the operating window of a unit and the operating
highest investment costs but the flexibility does not pay off.
envelope of the whole setup. The axes of the diagram represent
Thus, this setup should not be used at all.
the percentage of design capacity. 100 % is equal to the design
capacity of 10 t a1. With the help of these diagrams, it is possible to identify capacity bottlenecks.
4.2 Evaluation of Expansion Flexibility
The fermenter has a relatively low digression exponent of
0.51 (see Tab. 1) and the highest share in investment costs (see
The expansion flexibilities of the three setups presented in
Fig. 2). This means that the use of larger fermenter modules
Tab. 3 were evaluated for two different expansion scenarios
www.cet-journal.com
Plant design
339
www.cet-journal.com
340
G. Schembecker et al.
Scenario 1
Scenario 2
EFRogalski
EFJacob
EFRogalski
EFJacob
1.5
0.2
2.2
0.7
0.7
1.0
0.8
1.1
0.6
1.1
0.7
the higher investment costs of these setups compared to setup 1 and lead to a higher NPV. In the evaluation method
according to Rogalski, this benefit was considered as only the
expansion step itself was investigated. In this case, the lower
capacity limit of setups 2 and 3 is a major advantage, and the
financial benefits generated until expansion is needed can be
fully used.
EF of setup 1 is much higher in scenario 2 than in scenario 1.
The reasons for this are the additional investment costs for the
expansion that arise from the available module size. In both
scenarios, one 40-m3 fermenter is needed to reach the new
design capacities. While in scenario 2 this fermenter size fits
well to the new design capacity, in scenario 1 the overdesign of
this fermenter is too large, leading to high costs for the expansion. Thus, the difference in EAA between E and N of setup 1
in scenario 1 increases by only $ 0.3 million a1. In the case of the
other two setups, this increase is in the range of $ 1.3 million a1.
Consequently, in scenario 1, most of the additional income is
needed to pay for the expansion. In scenario 2, the economies
in scale of setup 1 are higher as compared to the other setups,
due to the larger fermenters. Therefore, parts of the profit generated by the time advantage of setups 2 and 3 are compensated for by the lower investment costs of setup 1. This trend
will be even more pronounced in scenarios with expansions to
even higher capacities.
To sum up, the results calculated with the method of Jacob
show the interaction between the demand development and
the apparatus sizes used. The adapted method allows a comparison between different market situations and enables the
decision for the setup with the highest expansion flexibility
with regard to long-run profit. In the given example, setup 2
with the higher numbering up should be preferred in the
expansion scenarios up to a doubling of the design capacity. In
scenarios with higher final capacities, setup 1 should be preferred as oversizing of the fermenter reduces the additional
investment costs required for further capacity increases.
5
The results of the EF evaluation can be seen in Tab. 6. The
calculated EAA are shown in Tab. 7. In contrast to the EF analysis using the method of Rogalski, setups 2 and 3 have a higher
expansion flexibility than setup 1 in both market scenarios.
This means that these two setups allow a more precise adaption to the market development than setup 1. One reason is
that production can be started earlier with setups 2 and 3, due
to the lower bound of the operating window compared to setup 1. The earnings made by this earlier start compensate for
Table 7. EAA (in $ million a1) for the different scenarios.
Setup
Scenario 1
Scenario 2
EAA(N)
EAA(E)
EAA(I)
EAA(N)
EAA(E)
10.4
10.7
12.1
13.3
16.2
10.0
11.5
12.1
12.8
16.4
10.0
11.2
12.1
12.8
16.1
www.cet-journal.com
Fields of Application
Plant design
341
Acknowledgment
The research leading to the results presented in this paper has
received funding from the European Communitys 7th Framework Program under grant agreement no. 228867, F3 Factory.
Three different methods for the evaluation of capacity flexibility of modular plants were presented. The first method allows
the investigation of the volume flexibility by analyzing the
operating range of each apparatus within the plant. Capacity
bottlenecks can be identified and additional costs for increasing the volume flexibility can be calculated. This makes the
presented method for volume flexibility evaluation very useful
for the design of modular plants, e.g., in seasonally volatile
markets. The evaluation methods according to Rogalski and
Jacob are related to expansion flexibility and help to integrate
future expansion scenarios in modular plant design. The method adapted from Rogalski [17] can be applied as an extension
for assessment with volume flexibility. The method adapted
from Jacob [3, 16] describes the long-term profit of an expansion scenario with the help of a DCF calculation and is especially useful in scenarios with strongly growing markets where
expansion can generate a significant economic benefit.
In future works, the methods presented should be used to
optimize plant setups concerning costs and flexibility. In this
way, a coupling with the process design phase can be achieved
by integrating the flexibility analysis into the process simula-
Symbols used
BEC
C
Cmax
Cmin
CRef
EAA
EFJacob
[$]
[t a1]
[t a1]
[t a1]
[$]
[$ a1]
[]
EFRogalski []
n
NPV
P
r
S
[]
[$]
[$]
[]
[m3, m2]
www.cet-journal.com
342
Sref
t
VF
G. Schembecker et al.
[m3, m2]
[a]
[]
reference size
time
volume flexibility
References
[1] H. P. Wiendahl, Planung modularer Fabriken, Hanser,
Munich 2005.
[2] D. Upton, J. Oper. Manag. 1995, 12, 205.
[3] A. Sethi, S. Sethi, Int. J. Flex. Manuf. Syst. 1990, 2 (4), 298.
[4] S. Buchholz, Chem. Eng. Process. Process Int. 2010, 49 (10),
993.
[5] D. Gerwin, Int. J. Oper. Prod. Manag. 1987, 7 (1), 38.
[6] J. M. Cheng, J. E. Simmons, M. Ritchie, J. Manuf. Technol.
Manag. 1997, 8 (3), 147.
[7] R. P. Parker, A. Wirth, Eur. J. Oper. Res. 1999, 118 (3), 429.
[8] R. E. Swaney, E. Grossmann, AIChE J. 1985, 31 (4), 621.
[9] R. E. Swaney, E. Grossmann, AIChE J. 1985, 31 (4), 631.
[10] I. E. Grossmann, C. A. Floudas, Comput. Chem. Eng. 1987,
11 (6), 675.
[11] C. A. Floudas, Z. H. Gms, M. G. Ierapetritou, Ind. Eng.
Chem. Res. 2001, 40 (20), 4267.
[12] D. A. Straub, I. E. Grossmann, Design Optimization of Stochastic Flexibility, Carnegie Mellon University Engineering
Design Research Center, Pittsburgh 1992.
[13] D. A. Straub, I. E. Grossmann, Integrated Statistical Metric of
Flexibility for Systems with Discrete State and Continuous
Parameter Uncertainties, Carnegie Mellon University Engineering Design Research Center, Pittsburgh 1989.
[14] F. V. Lima, Z. Jia, M. Ierapetritou, C. Georgakis, AIChE J.
2010, 56 (3), 702.
www.cet-journal.com
[15] A. Malcolm, J. Polan, L. Zhang, B. A. Ogunnaike, A. A. Linninger, AIChE J. 2007, 53 (8), 2048.
[16] H. Jacob, Z. Betriebswirtsch. 1974, 44, 299.
[17] S. Rogalski, Flexibility Measurement in Production Systems:
Handling Uncertainties in Industrial Production, Springer,
Berlin 2011.
[18] J. Oldenburg, M. Schlegel, J. Ulrich, T.-L. Hong, B. Krepinsky, G. Grossmann, A. Polt, H. Terhorst, in 17th European
Symposium on Computer Aided Process Engineering (Eds:
V. Plesu, P. S. Agachi), Elsevier, Amsterdam 2007.
[19] A. Wiesner, M. Schlegel, J. Oldenburg, L. Wrth, R. Hannmann, A. Polt, in 18th European Symposium on Computer
Aided Process Engineering (Eds: B. Braunschweig, X. Joulia),
Elsevier, Amsterdam 2008.
[20] J. R. Coleman, R. York, Ind. Eng. Chem. 1964, 56 (1), 28.
[21] S. Lier, M. Grnewald, Chem. Eng. Technol. 2011, 34 (5), 809.
[22] S. Lier, D. Wrsdrfer, M. Grnewald, Chem. Ing. Tech.
2012, 84 (12), 2164.
[23] T. Seifert, S. Sievers, C. Bramsiepe, G. Schembecker, Chem.
Eng. Process. 2012, 52, 140.
[24] M. Cheryan, Ultrafiltration Handbook, Technomic Publishing
Co., Basel 1986.
[25] D. R. Woods, Rules of Thumb in Engineering Practice, WileyVCH, Weinheim 2007.
[26] M. Baerns et al., Technische Chemie, Wiley-VCH, Weinheim
2008.
[27] C. S. Park, Contemporary Engineering Economics, Pearson
Education, New York 2010.
[28] S. B. Thakore, B. I Bhatt, Introduction to Process Engineering
and Design, Tata McGraw-Hill Publ. Co. Ltd., New Delhi
2007.