Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

On the Analysis and Fault-Diagnosis Tools for Small-Scale Heat and Power Plants

Arriagada, Jaime LU (2003)
Abstract
The deregulation of the electricity market drives utilities and independent power producers to operate heat and power plants as profit centers. In order to keep the economic margins on the credit side, the preferred measures have been to improve the electrical efficiency through changes in the hardware and boost the overall efficiency through e.g. combined heat and power (CHP) generation. The better understanding of global environmental issues is also pushing the development toward more advanced power plant technology that at the introduction stage may represent a risky option for the plant owner. Recently, there is a growing interest in improving the plant operation instead, and therefore the focus has been put on aspects related to the... (More)
The deregulation of the electricity market drives utilities and independent power producers to operate heat and power plants as profit centers. In order to keep the economic margins on the credit side, the preferred measures have been to improve the electrical efficiency through changes in the hardware and boost the overall efficiency through e.g. combined heat and power (CHP) generation. The better understanding of global environmental issues is also pushing the development toward more advanced power plant technology that at the introduction stage may represent a risky option for the plant owner. Recently, there is a growing interest in improving the plant operation instead, and therefore the focus has been put on aspects related to the RAM (reliability-availability-maintenance) of the plants.



Small- and mid-scale CHP plants, especially natural gas- and biomass-fueled, have been identified to be important to satisfy the needs of the energy market and help to mitigate the environmental factors in the short- and middle-term. One of the major challenges that these types of plants will face is attaining good RAM at the same time that they cannot support big O&M costs and a lot of personnel. Therefore the implementation of cheap and reliable IT-based tools that help to achieve this goal is essential.



Most power plants today are equipped with modern distributed control systems that through a considerable number of sensors deliver large amounts of data to the control room. This paves the way to the introduction of intelligent tools -derived from the artificial intelligence technology- such as artificial neural networks (ANNs) and genetic algorithms (GA). ANNs have a learning ability that makes them useful for the construction of powerful non-physical models based on data from the process, while GA has shown to be a robust optimization method based on the principle of the “survival-of-the-fittest”. Principally ANNs, but also to a lesser extent GA, have been applied to different case studies in this thesis, either alone or in conjunction with heat and mass balance programs (HMBPs) -the state-of-the-art tool in the field today. This thesis presents both theoretical and experimental case studies which have been validated with data from simulations and real plants, respectively. The studied tasks include design, optimization, system identification, and fault diagnosis of small- and medium-size heat and power plants. Some results of these case studies are powerful hybrid models that speed up calculations and fault diagnosis systems capable of recognizing developing faults and delivering early warnings to the plant operator. (Less)
Abstract (Swedish)
Popular Abstract in Swedish

Avregleringen av elmarknaden uppmuntrar kraftbolag och oberoende kraftproducenter till att driva sina kraftverksanläggningar på ett marknadsorienterat sätt. För att behålla de ekonomiska marginalerna på plussidan så har de mest populära åtgärderna varit att förbättra elverkningsgraden genom ändringar i hårdvaran samt höja totalverkningsgraden genom till exempel samproduktion av kraft och värme (kraftvärme). En bättre förståelse av globala miljöproblem är ytterligare en faktor som bidrar till utvecklingen av mer avancerad kraftverksteknologi, vilken åtminstone vid ett introduktionsstadium, kan utgöra ett riskabelt alternativ för kraftverksägaren. Under den senaste tiden har intresset för... (More)
Popular Abstract in Swedish

Avregleringen av elmarknaden uppmuntrar kraftbolag och oberoende kraftproducenter till att driva sina kraftverksanläggningar på ett marknadsorienterat sätt. För att behålla de ekonomiska marginalerna på plussidan så har de mest populära åtgärderna varit att förbättra elverkningsgraden genom ändringar i hårdvaran samt höja totalverkningsgraden genom till exempel samproduktion av kraft och värme (kraftvärme). En bättre förståelse av globala miljöproblem är ytterligare en faktor som bidrar till utvecklingen av mer avancerad kraftverksteknologi, vilken åtminstone vid ett introduktionsstadium, kan utgöra ett riskabelt alternativ för kraftverksägaren. Under den senaste tiden har intresset för förbättrad drift ökat och man fokuserar mer på aspekter relaterade till kraftverkens tillförlitlighet, tillgänglighet och underhåll (TTU).



Små- och medelskaliga kraftvärmeanläggningar, speciellt dessa som använder naturgas och biomassa som bränsle, har identifierats som viktiga alternativ för att uppfylla energimarknadens behov och för att lindra miljöaspekterna på kort sikt. En av de största utmaningarna som dessa anläggningar kommer att utsättas för är att de måste erhålla bra TTU samtidigt som de inte kan bära stora drift- och underhållskostnader och stora personalstyrkor. Därför är det viktigt att implementera billiga och pålitliga IT-baserade verktyg som hjälper att uppnå dessa målsättningar.



De flesta kraftverk utrustas numera med distribuerade reglersystem som skickar stora mängder information till kontrollrummet från ett stort antal mätpunkter. Detta banar väg för introduktion av så kallade intelligenta verktyg (som härstammar från området artificiell intelligens) som till exempel artificiella neurala nätverk (ANN) och genetiska algoritmer (GA). ANN har inlärningsförmåga vilket gör dem användbara för att bygga icke-fysikaliska modeller baserade på data från processen, medan GA har visat sig vara en robust optimeringsmetod baserad på principen ”survival-of-the-fittest”. I den här avhandlingen, tillämpas först och främst ANN, men också i mindre omfattning GA, i olika studier. Dessa verktyg appliceras antingen enskilt eller i kombination med värmebalansprogram (VBP) – ett state-of-the-art verktyg inom området. Den här avhandlingen presenterar teoretiska och experimentella studier, som har validerats med data från både simuleringar och verkliga anläggningar. De här studierna inkluderar design, optimering, systemidentifiering och feldiagnostik av små- och medelskaliga kraftvärmeverk. Några resultat från de här studierna är kraftfulla hybridmodeller som förkortar beräkningstiderna samt feldiagnostiksystem som är kapabla att identifiera icke fullt utvecklade fel och som dessutom har möjlighet att leverera varningssignaler till operatören. (Less)
Please use this url to cite or link to this publication:
author
supervisor
opponent
  • Dr Karlsson, Agne, Finspång
organization
publishing date
type
Thesis
publication status
published
subject
keywords
Thermal engineering, applied thermodynamics, Termisk teknik, termodynamik, genetic algorithms, neural networks, fault diagnosis, maintenance, availability, reliability, energy analysis, combined heat and power, Methods and tools, small scale
pages
228 pages
publisher
Department of Heat and Power Engineering, Lund university
defense location
Room M:B of the M-building, Lund Institute of Technology
defense date
2003-12-15 10:15:00
external identifiers
  • other:ISRN:LUTMDN/TMHP--03/1015--SE
ISBN
91-628-5843-2
language
English
LU publication?
yes
additional info
Article: 1. Arriagada, J., Assadi, M., (2000): Air Bottoming Cycle for Gas Turbines. ISME-2000, Mechanical Engineering Conference, Teheran, Iran. Article: 2. Arriagada, J., Rosén, P., Torisson, T., (2001): A novel gas turbine concept for combined power, heat and cooling generation. ASME Turbo Expo 2001, New Orleans, USA. Article: 3. Arriagada, J., Azimian, A., Assadi, M., (2004): Generation of Steam Tables Using Artificial Neural Networks. Accepted for publication in Heat Transfer Engineering – An International Journal, Taylor and Francis Ltd, Vol. 25(2), February 2004. Article: 4. Assadi, M., Mesbahi, E., Torisson, T., Lindquist, T., Arriagada, J., Olausson, P., (2001): A Novel Correction Technique for Simple Gas Turbine Parameters. ASME Turbo Expo 2001, New Orleans, USA. Article: 5. Arriagada, J., Olausson, P., Selimovic, A., (2002): Artificial Neural Network Simulator for SOFC Performance Prediction. Journal of Power Sources, Elsevier Science, Vol. 112, pp. 54-60. Article: 6. Arriagada, J., Costantini, M., Olausson, P., Assadi, M., Torisson, T., (2003): Artificial Neural Network model for a Biomass-fueled Boiler. ASME Turbo Expo 2003, Atlanta, USA. Article: 7. Olausson, P., Häggståhl, D., Arriagada, J., Dahlquist, E., Assadi, M., (2003): Hybrid Model of an Evaporative Gas Turbine Power Plant Utilizing Physical Models and Artificial Neural Networks. ASME Turbo Expo - 2003, Atlanta, USA. Article: 8. Mesbahi, E., Arriagada, J., Assadi, M., Ghorban, H. (2003): Diesel Engine Fault Diagnosis by Means of ANNs: A Comparison Between Fault Pattern and Residual Method. Submitted to Journal of Marine Design and Operations. Article: 9. Arriagada, J., Genrup, M., Loberg, A., Assadi, M., (2003): Fault Diagnosis System for anIndustrial Gas Turbine by Means of Neural Networks. IGTC2003, International Gas Turbine Congress, Tokyo, Japan. Article: 10. Fredriksson-Möller, B., Arriagada, J., Assadi, M., Potts, I. (2003): Optimization of a GT/SOFC System with CO2 Capture. Submitted to Journal Power Sources, Elsevier Science.
id
c7e86533-260d-40f6-9798-53522107158d (old id 466507)
date added to LUP
2016-04-01 16:54:45
date last changed
2018-11-21 20:45:10
@phdthesis{c7e86533-260d-40f6-9798-53522107158d,
  abstract     = {{The deregulation of the electricity market drives utilities and independent power producers to operate heat and power plants as profit centers. In order to keep the economic margins on the credit side, the preferred measures have been to improve the electrical efficiency through changes in the hardware and boost the overall efficiency through e.g. combined heat and power (CHP) generation. The better understanding of global environmental issues is also pushing the development toward more advanced power plant technology that at the introduction stage may represent a risky option for the plant owner. Recently, there is a growing interest in improving the plant operation instead, and therefore the focus has been put on aspects related to the RAM (reliability-availability-maintenance) of the plants.<br/><br>
<br/><br>
Small- and mid-scale CHP plants, especially natural gas- and biomass-fueled, have been identified to be important to satisfy the needs of the energy market and help to mitigate the environmental factors in the short- and middle-term. One of the major challenges that these types of plants will face is attaining good RAM at the same time that they cannot support big O&amp;M costs and a lot of personnel. Therefore the implementation of cheap and reliable IT-based tools that help to achieve this goal is essential.<br/><br>
<br/><br>
Most power plants today are equipped with modern distributed control systems that through a considerable number of sensors deliver large amounts of data to the control room. This paves the way to the introduction of intelligent tools -derived from the artificial intelligence technology- such as artificial neural networks (ANNs) and genetic algorithms (GA). ANNs have a learning ability that makes them useful for the construction of powerful non-physical models based on data from the process, while GA has shown to be a robust optimization method based on the principle of the “survival-of-the-fittest”. Principally ANNs, but also to a lesser extent GA, have been applied to different case studies in this thesis, either alone or in conjunction with heat and mass balance programs (HMBPs) -the state-of-the-art tool in the field today. This thesis presents both theoretical and experimental case studies which have been validated with data from simulations and real plants, respectively. The studied tasks include design, optimization, system identification, and fault diagnosis of small- and medium-size heat and power plants. Some results of these case studies are powerful hybrid models that speed up calculations and fault diagnosis systems capable of recognizing developing faults and delivering early warnings to the plant operator.}},
  author       = {{Arriagada, Jaime}},
  isbn         = {{91-628-5843-2}},
  keywords     = {{Thermal engineering; applied thermodynamics; Termisk teknik; termodynamik; genetic algorithms; neural networks; fault diagnosis; maintenance; availability; reliability; energy analysis; combined heat and power; Methods and tools; small scale}},
  language     = {{eng}},
  publisher    = {{Department of Heat and Power Engineering, Lund university}},
  school       = {{Lund University}},
  title        = {{On the Analysis and Fault-Diagnosis Tools for Small-Scale Heat and Power Plants}},
  year         = {{2003}},
}