A Brief Review of Methods For The Design and Synthesis Optimization of Energy Systems (#76638) - 65612
A Brief Review of Methods For The Design and Synthesis Optimization of Energy Systems (#76638) - 65612
A Brief Review of Methods For The Design and Synthesis Optimization of Energy Systems (#76638) - 65612
Christos A. FRANGOPOULOS*
National Technical University of Athens
Dept. of Naval Architecture and Marine Engineering
Heroon Polytechniou 9, 157 73 Zografou, Greece.
Tel: +30-210-7721108, Fax: +30-210-7721117
E-mail: [email protected]
Michael R. von SPAKOVSKY
Virginia Polytechnic Institute and State University
Mechanical Engineering Department,
Blacksburg, VA 24061, USA.
Tel: +1-540-2316684, Fax: +1-540-2319100,
E-mail: [email protected]
Enrico SCIUBBA
Università di Roma 1 “ La Sapienza”
Dipartimento di Meccanica ed Aeronautica
Via Eudossiana 18, 00185 Roma, Italy.
Tel: +39-06-44585244, Fax: +39-06-4881759
E-mail: [email protected]
Abstract
The optimization of energy systems is of crucial importance for a rational use of natural
and economic resources and for minimizing their adverse effects on the environment.
Optimizing such systems may be considered at three levels: synthesis (configuration),
design (component characteristics), and operation. The first two of these levels are ex-
amined in this article. After a discussion on the uniqueness of the solution and the pos-
sibility of finding this solution, the principal approaches and methods for solving the
optimization problem are described in brief.
Key words: thermoeconomics, design optimization of energy systems, synthesis
optimization of energy systems, artificial ıntelligence, genetic algorithms,
mathematical programming, decomposition
For example, the targets of class (b) can serve as Furthermore, if the system is completely
heuristics or rules in class (a) and they can be specified (both z and w are given), then an opera-
embedded in the optimization procedures of class tion optimization problem is indicated:
(c) to the benefit of the whole process.
min imize Fop ( x ) (1) op
x
4. Mathematical Statement of the Complete
Optimization Problem 5. Representative Methods for the Solution of
The objective function of the complete op- the Synthesis Optimization Problem
timization problem (i.e. synthesis, design, and The design optimization problem can be
operation) is written in the general form: solved by a number of methods described in the
min imize F(x, w, z ) (1) literature (Floudas 1995, Bejan et al. 1996,
x ,w,z Stoecker 1989, Rao 1996, Reklaitis et al. 1983,
subject to the constraints Papalambros and Wilde 2000, Moré and Wright
u12 u12
u21 u21 w 22
w 21
unit–level
C* = C1∗ + C2∗ optimum
System-
system–level level ORS w 2 = {w 21 , w 22 }
optimum u12
z 2 = {z 21 , z 22 }
(ψ ∗
, ξ ∗ , C ∗∗ )
u21
Figure 3. Local (unit-level) and global (system-level) optimizations for LGO and ILGO
Of course, as with the other decomposition ap- Expert Systems are based on relational
proaches, the principle disadvantage of LGO is languages that use the symbolism of formal pro-
that it is very computationally intensive. To cir- positional logic. They draw inferences from a
cumvent this, ILGO instead of LGO may be ap- number of facts stored in a particular database,
plied since the former avoids the need for creat- properly called a knowledge base. These facts
ing any of the ORSs and avoids as well the nest- can be design data, design rules, physical or logi-
ing inherent in the other decomposition ap- cal constraints, etc. Each ES manipulates this
proaches. ILGO accomplishes the former by us- knowledge in its own way, according to a logical
ing derivative information in the form of what procedure contained in its inference engine.
are called shadow prices (derivatives of the op- Space limitations do not allow a detailed presen-
timal value of a function with respect to certain tation of AI and ES techniques, but the interested
variables) to intelligently move along the system- reader can find information in the literature
level ORS towards the system-level optimum, (Sciubba and Melli 1998, Sriram 1997, Green
and it accomplishes the latter by incorporating 1992).
system-level information directly into the unit-
level objectives through the use of shadow prices 6. Closure
5.7. Artificial Intelligence and Expert Systems As one goes from operation optimization to
techniques design and synthesis optimization of energy sys-
tems, the problem becomes much more difficult
In the preceding, it is tacitly assumed that not only from a computational but also from a
all process design calculations can be carried out methodological point of view. Several methods
by properly implemented automated routines. have been developed, which have been shown to
Process design is a highly labor intensive and be successful at least for particular classes of
highly interdisciplinary task and is, therefore, by problems, even when the number of degrees of
necessity performed by a team of specialists: as a freedom is large and the models involved highly
consequence, it is also very expensive in mone- nonlinear and complex. However, the effort to
tary terms, and there is a strong incentive to re- develop even more capable and more generally
duce this labor intensity (measured in man- applicable methods continues.
hours). The only task that has as of yet not been
fully automated is the conceptual one: the choice References
of the type and of the characteristics of the proc-
ess itself. This automation can be implemented Bausa, J., and Tsatsaronis, G., 2001, “Dy-
by a direct application of the very powerful Arti- namic Optimization of Startup and Load-
ficial Intelligence (AI) techniques, whose spe- increasing Processes in Power Plants – Part I:
cific task is to allow the codification of proce- Method, Part II: Application,” ASME Journal of
dures that somehow mimic the thinking patterns Engineering for Gas Turbines and Power, Vol.
of the human mind. Currently, only a subset of 123, pp. 246-254.
these techniques, called Expert Systems (ES), Bejan, A., Tsatsaronis, G., and Moran, M.,
have been successfully applied to energy sys- 1996, Thermal Design and Optimization, John
tems. ESs can be used to reproduce the engi- Wiley & Sons, New York.
neer’s decisional path that proceeds from the
Bondy, J. A., and Murty, U.S.R., 1976,
design data and constraints to possible process
Graph Theory with Applications, Elsevier.
configurations.
Frangopoulos, C. A., Lygeros, A. I., Markou, C. Munoz, J. R., and von Spakovsky M. R., 2001b,
“The Application of Decomposition to the Large
T. and Kaloritis P., 1996, “Thermo-economic
Operation Optimization of the Hellenic Aspro- Scale Synthesis / Design Optimization of Air-
pyrgos Refinery Combined-cycle Cogeneration craft Energy Systems, International Journal of
Applied Thermodynamics, Vol. 4, No. 2, pp. 61-
System,” Applied Thermal Engineering, Vol. 16
No. 12, pp. 949-958. 76.
Gen, M., and Cheng, R., 1997, Genetic Algo- Olsommer, B., von Spakovsky, M. R., and
rithms and Engineering Design, J. Wiley and Favrat, D., 1999, “An Approach for Time-
dependent Thermoeconomic Modeling and Op-
Sons, New York.
timization of Energy System Synthesis, Design
Goldberg, D. E., 1989, Genetic Algorithms in and Operation: Part I – Methodology and Re-
Search, Optimization and Machine Learning, sults, Part II – Reliability and Availability,” In-
Addison-Wesley Publishing Co., Inc., Reading, ternational Journal of Applied Thermo-
Massachusetts. dynamics, Vol. 2 No. 3, 1999, pp. 97-114, and
Green, M., ed., 1992, Knowledge Aided Design, Vol. 2 No. 4, pp. 177-186.
Academic Press. Papalambros, P. Y., and Wilde, J., 2000, Princi-
Kott, A. S., May, J. H. and Hwang, C. C., 1989, ples of Optimal Design – Modeling and Compu-
“An Autonomous Artificial Designer of Thermal tation, Cambridge University Press, Cambridge,
Energy Systems: Part 1 – Theoretical Considera- U.K.
tions, Part 2 – Solution Algorithm,” ASME Jour- Rao, S. S., 1996, Engineering Optimization:
nal of Engineering for Gas Turbines and Power, Theory and Practice, 3rd ed., J. Wiley & Sons,
Vol. 111, pp. 728-739. Inc., New York.
Linial, N., Lovasz, L., and Widgerson, A., 1986, Reklaitis, G. V., Ravindran, A., and Ragsdell, K.
“A Geometric Interpretation of Graph Connec- M., 1983, Engineering Optimization – Methods
tivity and its Algorithmic Applications,” 27th and Applications, J. Wiley & Sons, New York.
Symposium on Foundations of Computer Sci-
ence, pp. 39-48. Sciubba, E., 1998, “Toward Automatic Process
Simulators: Part I – Modular Numerical Proce-
Linnhoff, B., et al., 1982, A User Guide on Proc- dures, Part II – An Expert System for Process
ess Integration for the Efficient Use of Energy, Synthesis,” ASME Journal of Engineering for
Institution of Chemical Engineers, Rugby, Gas Turbines and Power Vol. 120, pp. 1-16.
Warks, England.
Sciubba, E., and Melli, R., 1998, Artificial Intel-
Linnhoff, B., 1989, “Pinch Technology for the ligence in Thermal Systems Design: Concepts
Synthesis of Heat and Power Systems,” J. of En- and Applications, Nova Science Publishers, Inc.,
ergy Resources Technology, Vol. 111, pp. 137- Commack, New York.
148.