Modelling Towards Control of Dynamic Systems : Applications on RDF Fired CFB Performance and DHN Distribution

(1)

ISBN 978-91-7485-475-6 ISSN 1651-4238

Address: P.O. Box 883, SE-721 23 Västerås. Sweden Address: P.O. Box 325, SE-631 05 Eskilstuna. Sweden E-mail: info@mdh.se Web: www.mdh.se

Dynamic Systems

Applications on RDF Fired CFB Performance and DHN

Distribution

Nathan Zimmerman mer m a n M O D EL LIN G T O W A R D S C O N TR O L O F D Y N A M IC S YS TE M S 2020

(2)

Mälardalen University Press Dissertations No. 319

MODELLING TOWARDS CONTROL OF DYNAMIC SYSTEMS

APPLICATIONS ON RDF FIRED CFB PERFORMANCE AND DHN DISTRIBUTION

Nathan Zimmerman 2020

School of Business, Society and Engineering

MODELLING TOWARDS CONTROL OF DYNAMIC SYSTEMS

APPLICATIONS ON RDF FIRED CFB PERFORMANCE AND DHN DISTRIBUTION

Nathan Zimmerman 2020

(3)

ISSN 1651-4238

Printed by E-Print AB, Stockholm, Sweden

ISSN 1651-4238

(4)

MODELLING TOWARDS CONTROL OF DYNAMIC SYSTEMS

APPLICATIONS ON RDF FIRED CFB PERFORMANCE AND DHN DISTRIBUTION

Nathan Zimmerman

Akademisk avhandling

som för avläggande av teknologie doktorsexamen i energi- och miljöteknik vid Akademin för ekonomi, samhälle och teknik kommer att offentligen försvaras tisdagen den 29 september 2020, 09.15 i Delta, Mälardalens högskola, Västerås.

Fakultetsopponent: Professor Natasa Nord, Norwegian University of Science and Technology

Akademin för ekonomi, samhälle och teknik

MODELLING TOWARDS CONTROL OF DYNAMIC SYSTEMS

APPLICATIONS ON RDF FIRED CFB PERFORMANCE AND DHN DISTRIBUTION

Nathan Zimmerman

Akademisk avhandling

som för avläggande av teknologie doktorsexamen i energi- och miljöteknik vid Akademin för ekonomi, samhälle och teknik kommer att offentligen försvaras tisdagen den 29 september 2020, 09.15 i Delta, Mälardalens högskola, Västerås.

Fakultetsopponent: Professor Natasa Nord, Norwegian University of Science and Technology

(5)

Abstract

The combination of global warming along with increasing energy demand necessitates the importance of improving processes pertaining to the production and consumption of energy in combined heat and power plants. This thesis brings to light transient factors currently burdening process performance for circulating fluidized bed boilers (CFBs) combusting refuse derived fuels (RDFs) and district heating networks (DHN). These two domains are not completely disconnected from one another, which is the case for Northern European countries. Heat can be generated from a central location to be distributed through a network of customers to meet a heating demand. Results show that first-principle modelling techniques have the capacity to capture transients factors associated within the aforementioned entwined energy systems.

On the production side, obtaining real-time information pertaining to the lower heating value of refuse derived fuel affords the ability to implement feed-forward model predictive control. Therefore, feed-forward model predictive control has the potential to minimize combustion temperature swings by making the necessary controls moves before changes in the fuel’s composition are actualized by the process. On the consumption side, attaining a deeper understanding of district heating network dynamics, e.g. heat propagation, network losses, distribution delays, and end-user requirements, introduces the possibility to analyse network performance and reduce peak load production. The perspective of quick network performance can be achieved by an automated approach to building and simulating district heating networks. Nonconventional end-user heating configurations, e.g. homes utilizing district heating and a heat pump, has the potential of illustrating how heating consumption patterns may change over time. Peak load reduction is achievable in district heating networks when it is possible to reduce network supply temperature. This can be achieved by predicting end-user heating requirements and using this information for feed-forward model predictive control.

The overall observations made in this thesis demonstrates that process improvements are obtainable for transient energy systems. Despite the presented work focusing on only one type of energy production and one type of consumption, the approach described unlocks a flexibility that eliminates the need for unambiguous modelling and simulations by allowing for the reusability of model components. The exportability of these models further distinguishes them, as they can be used to test new control approaches within an energy system as real-time predictions within each energy sub-system become more accessible.

ISBN 978-91-7485-475-6 ISSN 1651-4238

Abstract

The combination of global warming along with increasing energy demand necessitates the importance of improving processes pertaining to the production and consumption of energy in combined heat and power plants. This thesis brings to light transient factors currently burdening process performance for circulating fluidized bed boilers (CFBs) combusting refuse derived fuels (RDFs) and district heating networks (DHN). These two domains are not completely disconnected from one another, which is the case for Northern European countries. Heat can be generated from a central location to be distributed through a network of customers to meet a heating demand. Results show that first-principle modelling techniques have the capacity to capture transients factors associated within the aforementioned entwined energy systems.

On the production side, obtaining real-time information pertaining to the lower heating value of refuse derived fuel affords the ability to implement feed-forward model predictive control. Therefore, feed-forward model predictive control has the potential to minimize combustion temperature swings by making the necessary controls moves before changes in the fuel’s composition are actualized by the process. On the consumption side, attaining a deeper understanding of district heating network dynamics, e.g. heat propagation, network losses, distribution delays, and end-user requirements, introduces the possibility to analyse network performance and reduce peak load production. The perspective of quick network performance can be achieved by an automated approach to building and simulating district heating networks. Nonconventional end-user heating configurations, e.g. homes utilizing district heating and a heat pump, has the potential of illustrating how heating consumption patterns may change over time. Peak load reduction is achievable in district heating networks when it is possible to reduce network supply temperature. This can be achieved by predicting end-user heating requirements and using this information for feed-forward model predictive control.

The overall observations made in this thesis demonstrates that process improvements are obtainable for transient energy systems. Despite the presented work focusing on only one type of energy production and one type of consumption, the approach described unlocks a flexibility that eliminates the need for unambiguous modelling and simulations by allowing for the reusability of model components. The exportability of these models further distinguishes them, as they can be used to test new control approaches within an energy system as real-time predictions within each energy sub-system become more accessible.

ISBN 978-91-7485-475-6 ISSN 1651-4238

(6)

(7)

"Even if you fall on your face, you’re still

moving forward."

- Victor Kiam

"Even if you fall on your face, you’re still

moving forward."

- Victor Kiam

(8)

Acknowledgements

First and foremost I would like to thank my parents for their endless support and guidance. If it was not for them I would never have made the leap to study in Sweden, which has profoundly impacted my life. I also owe a debt of gratitude to my Sambo Barrett. Without her unyielding patience and support through the duration of my PhD I would have certainly fallen to pieces. She is my foundation and the being of my happiness. I also want to deeply thank my supervisory team of Erik Dahlquist, Konstantinos Kyprianidis and Carl-Fredrik Lindberg, without whom I would have struggled with my research focus. The attentiveness they have been able to afford me with countless hours of mentorship and guidance has challenged me to work at a higher level. I would also like to thank my friends and colleagues, past and present, at Mälardalen University for the knowledge I have had the privilege of obtaining from them during my studies. Finally, I would like to thank the individuals I have worked with at Mälarenergi and their persistence regarding how coffee time (fika) is essential to developing personal and professional relationships.

i

Acknowledgements

First and foremost I would like to thank my parents for their endless support and guidance. If it was not for them I would never have made the leap to study in Sweden, which has profoundly impacted my life. I also owe a debt of gratitude to my Sambo Barrett. Without her unyielding patience and support through the duration of my PhD I would have certainly fallen to pieces. She is my foundation and the being of my happiness. I also want to deeply thank my supervisory team of Erik Dahlquist, Konstantinos Kyprianidis and Carl-Fredrik Lindberg, without whom I would have struggled with my research focus. The attentiveness they have been able to afford me with countless hours of mentorship and guidance has challenged me to work at a higher level. I would also like to thank my friends and colleagues, past and present, at Mälardalen University for the knowledge I have had the privilege of obtaining from them during my studies. Finally, I would like to thank the individuals I have worked with at Mälarenergi and their persistence regarding how coffee time (fika) is essential to developing personal and professional relationships.

(9)

Summary

The implications of global warming coupled with the future prospects of continuing increases in energy demands has led to a growing demand for more efficient energy production processes. When striving towards more efficient energy systems, it is important to analyse not only primary energy production but also the consumption. In this work, the primary production is reviewed with a focus on the combustion technology of circulating fluidized bed boil-ers. This type of boiler is commonly utilized in district heating networks in Northern European countries. Thus, the focus on the consumption side of this work is on district heating networks. Renewable solid fuels present an oppor-tunity to transition away from fossil-based fuels. This initial step alone can help in mitigating greenhouse gases, as renewable fuels can be categorized as contributing to a reduction in net greenhouse gas emissions. Moreover, fur-ther reductions can be made by achieving a greater understanding of district heating network dynamics and end-user heating requirements.

On the production side, a model of the combustion process incorporating the most important signals and parameters can be used in the design of an MPC controller (model predictive control) with feed-forward. This control strategy mitigates the temperature fluctuations within the combustion zones, which maintains the efficiency of the boiler and keeps emissions low in the event of disturbances. The disturbance of interest in this work is that of the fuel’s lower heating value.

The results presented in this thesis show that an MPC controller with feed-forward, applied to a fluidized bed boiler using waste as fuel, can minimize the disturbances caused by fluctuations in the fuel’s lower heating value. A dynamic model has been used in the design of the MPC controller, which is a simplified model validated against a dynamic physical model of the process.

On the consumption side, the focus is on reducing peak load production by reducing the supply temperature to district heating networks while achieving higher stability in the return temperature. A physical model of a large-scale district heating network has been developed by aggregating the outlying re-gions of the network into six distinct rere-gions in order to ascertain transient network dynamics. The model of the regions include, but are not limited to, the propagation of heat, network losses and distribution delays. From these three aspects alone, better understanding of the network dynamics can assist operators. One such approach is to automate the construction of models of dis-trict heating networks for fast network analysis (e.g. delays and temperature ii

Summary

The implications of global warming coupled with the future prospects of continuing increases in energy demands has led to a growing demand for more efficient energy production processes. When striving towards more efficient energy systems, it is important to analyse not only primary energy production but also the consumption. In this work, the primary production is reviewed with a focus on the combustion technology of circulating fluidized bed boil-ers. This type of boiler is commonly utilized in district heating networks in Northern European countries. Thus, the focus on the consumption side of this work is on district heating networks. Renewable solid fuels present an oppor-tunity to transition away from fossil-based fuels. This initial step alone can help in mitigating greenhouse gases, as renewable fuels can be categorized as contributing to a reduction in net greenhouse gas emissions. Moreover, fur-ther reductions can be made by achieving a greater understanding of district heating network dynamics and end-user heating requirements.

On the production side, a model of the combustion process incorporating the most important signals and parameters can be used in the design of an MPC controller (model predictive control) with feed-forward. This control strategy mitigates the temperature fluctuations within the combustion zones, which maintains the efficiency of the boiler and keeps emissions low in the event of disturbances. The disturbance of interest in this work is that of the fuel’s lower heating value.

The results presented in this thesis show that an MPC controller with feed-forward, applied to a fluidized bed boiler using waste as fuel, can minimize the disturbances caused by fluctuations in the fuel’s lower heating value. A dynamic model has been used in the design of the MPC controller, which is a simplified model validated against a dynamic physical model of the process.

On the consumption side, the focus is on reducing peak load production by reducing the supply temperature to district heating networks while achieving higher stability in the return temperature. A physical model of a large-scale district heating network has been developed by aggregating the outlying re-gions of the network into six distinct rere-gions in order to ascertain transient network dynamics. The model of the regions include, but are not limited to, the propagation of heat, network losses and distribution delays. From these three aspects alone, better understanding of the network dynamics can assist operators. One such approach is to automate the construction of models of dis-trict heating networks for fast network analysis (e.g. delays and temperature ii

(10)

propagation). Furthermore, historical network data and consumer consump-tion enable the estimaconsump-tion of a network’s heating requirements, i.e. end-user load. This information can be used in an MPC with feed-forward to regulate a network’s supply temperature. In a closed-loop simulation with a set-point on the return temperature, the MPC is able to take the appropriate control ac-tions to adjust the network’s mass flow rate. The results suggest that traditional network control strategies can be improved with feed-forward MPC and that peak-load heat production can also be further reduced.

The consumption requirements of end users connected to a district heating network will not remain constant in future. Heat pumps are gaining ground as a viable contender to district heating, but they can also be supplementary to a local heating network in the form of a hybrid heating system. The external influence of a heat pump, running in parallel with a home connected to dis-trict heating, provides an understanding of how disdis-trict heating consumption patterns may change over time. The results suggest that a heat pump can be useful in conjunction with district heating during the Spring and Autumn, pro-viding savings in the heating cost, while propro-viding cooling during the Summer months.

The present and future energy system needs to be reviewed holistically in order to reduce primary energy production and consumption and alleviate po-tential environmental ramifications. The development of modelling libraries based on first-principle modelling techniques affords a higher level of durabil-ity in energy system modelling. Although the work presented in this thesis fo-cuses on only one type of energy production and one type of consumption, the approach described in this thesis unlocks a flexibility that eliminates the need for unambiguous modelling and simulations by allowing for the reusability of model components. The exportability of these models further distinguishes them, as they can be used to test new control approaches within an energy sys-tem as real-time disturbance predictions within each subsyssys-tem of the energy system become more accessible.

iii

propagation). Furthermore, historical network data and consumer consump-tion enable the estimaconsump-tion of a network’s heating requirements, i.e. end-user load. This information can be used in an MPC with feed-forward to regulate a network’s supply temperature. In a closed-loop simulation with a set-point on the return temperature, the MPC is able to take the appropriate control ac-tions to adjust the network’s mass flow rate. The results suggest that traditional network control strategies can be improved with feed-forward MPC and that peak-load heat production can also be further reduced.

The consumption requirements of end users connected to a district heating network will not remain constant in future. Heat pumps are gaining ground as a viable contender to district heating, but they can also be supplementary to a local heating network in the form of a hybrid heating system. The external influence of a heat pump, running in parallel with a home connected to dis-trict heating, provides an understanding of how disdis-trict heating consumption patterns may change over time. The results suggest that a heat pump can be useful in conjunction with district heating during the Spring and Autumn, pro-viding savings in the heating cost, while propro-viding cooling during the Summer months.

The present and future energy system needs to be reviewed holistically in order to reduce primary energy production and consumption and alleviate po-tential environmental ramifications. The development of modelling libraries based on first-principle modelling techniques affords a higher level of durabil-ity in energy system modelling. Although the work presented in this thesis fo-cuses on only one type of energy production and one type of consumption, the approach described in this thesis unlocks a flexibility that eliminates the need for unambiguous modelling and simulations by allowing for the reusability of model components. The exportability of these models further distinguishes them, as they can be used to test new control approaches within an energy sys-tem as real-time disturbance predictions within each subsyssys-tem of the energy system become more accessible.

(11)

Sammanfattning

Den globala uppvärmningen i kombination med prognosen för fortsatt ökat energibehov har lett till ökad efterfrågan på effektivare energiproduktionspro-cesser. I strävan efter mer effektiva energisystem är det viktigt att analysera inte bara den primära energiproduktionen utan även konsumtionen. I detta arbete granskas primärproduktionen med fokus på förbränningstekniken för pannor med cirkulerad fluidiserad bädd. Denna typ av panna används ofta i fjärrvärmenät i Nordeuropeiska länder. På konsumtionssidan ligger fokus i detta arbete på fjärrvärmenätet. Förnybara fasta bränslen utgör en möjlighet att gå ifrån fossila bränslen. Enbart denna åtgärd hjälper till att minska växthus-gaserna eftersom förnybara bränslen bidrar till en nettominskning. Dessutom kan ytterligare minskningar göras med en större förståelse av fjärrvärmenätets dynamik och slutanvändarens värmebehov.

På produktionssidan kan en modell av förbränningsprocessen med de vik-tigaste signalerna och parametrarna användas i designen av en MPC regula-tor (model predictive control) med framkoppling. Denna reglerstrategi min-skar temperatursvängningarna i förbränningen vilket bibehåller effektiviteten i pannan och håller utsläppen låga vid störningar. Ett exempel på en störning är en förändring i bränslets värmevärde.

Resultaten som presenteras i avhandlingen visar att en MPC regulator med framkoppling, tillämpat på en panna med fluidiserad bädd som använder avfall som bränsle, kan minimera störningarna som värmevärdet i bränslet orsakar. En dynamisk modell har använts vid designen av MPC regulatorn, vilken är en förenklad modell validerad mot en dynamisk fysikalisk modell av processen.

På konsumtionssidan är fokus att minska topparna i produktionen samtidigt som stabiliteten i returtemperaturen förbättras. En fysikalisk modell av ett storskaligt fjärrvärmenät har utvecklats genom att slå ihop de yttre regionerna i nätet till sex distinkta regioner för att säkerställa dynamiken i nätet.

Modellerna av regionerna inkluderar men är inte begränsade till transport av värme, förluster i nätet och distributionsfördröjningar. Enbart dessa tre aspek-ter ger en bättre förståelse av dynamiken i nätet vilket kan hjälpa operatörerna. För att göra nätverksanalys snabbare av t.ex. fördröjningar och värmeutbred-ning kan konstruktionen av modellen av fjärrvärmenät automatiseras. Med hjälp av historiska data från fjärrvärmenätet och konsumenternas förbrukning kan dessutom nätets värmebehov, dvs. lasten hos slutanvändarna estimeras. Denna information kan användas i MPC med framkoppling för reglering av fjärrvärmenätets temperatur.

iv

Sammanfattning

Den globala uppvärmningen i kombination med prognosen för fortsatt ökat energibehov har lett till ökad efterfrågan på effektivare energiproduktionspro-cesser. I strävan efter mer effektiva energisystem är det viktigt att analysera inte bara den primära energiproduktionen utan även konsumtionen. I detta arbete granskas primärproduktionen med fokus på förbränningstekniken för pannor med cirkulerad fluidiserad bädd. Denna typ av panna används ofta i fjärrvärmenät i Nordeuropeiska länder. På konsumtionssidan ligger fokus i detta arbete på fjärrvärmenätet. Förnybara fasta bränslen utgör en möjlighet att gå ifrån fossila bränslen. Enbart denna åtgärd hjälper till att minska växthus-gaserna eftersom förnybara bränslen bidrar till en nettominskning. Dessutom kan ytterligare minskningar göras med en större förståelse av fjärrvärmenätets dynamik och slutanvändarens värmebehov.

På produktionssidan kan en modell av förbränningsprocessen med de vik-tigaste signalerna och parametrarna användas i designen av en MPC regula-tor (model predictive control) med framkoppling. Denna reglerstrategi min-skar temperatursvängningarna i förbränningen vilket bibehåller effektiviteten i pannan och håller utsläppen låga vid störningar. Ett exempel på en störning är en förändring i bränslets värmevärde.

Resultaten som presenteras i avhandlingen visar att en MPC regulator med framkoppling, tillämpat på en panna med fluidiserad bädd som använder avfall som bränsle, kan minimera störningarna som värmevärdet i bränslet orsakar. En dynamisk modell har använts vid designen av MPC regulatorn, vilken är en förenklad modell validerad mot en dynamisk fysikalisk modell av processen.

På konsumtionssidan är fokus att minska topparna i produktionen samtidigt som stabiliteten i returtemperaturen förbättras. En fysikalisk modell av ett storskaligt fjärrvärmenät har utvecklats genom att slå ihop de yttre regionerna i nätet till sex distinkta regioner för att säkerställa dynamiken i nätet.

Modellerna av regionerna inkluderar men är inte begränsade till transport av värme, förluster i nätet och distributionsfördröjningar. Enbart dessa tre aspek-ter ger en bättre förståelse av dynamiken i nätet vilket kan hjälpa operatörerna. För att göra nätverksanalys snabbare av t.ex. fördröjningar och värmeutbred-ning kan konstruktionen av modellen av fjärrvärmenät automatiseras. Med hjälp av historiska data från fjärrvärmenätet och konsumenternas förbrukning kan dessutom nätets värmebehov, dvs. lasten hos slutanvändarna estimeras. Denna information kan användas i MPC med framkoppling för reglering av fjärrvärmenätets temperatur.

(12)

I en simulering i sluten loop med börvärde på returtemperaturen kan MPC regulatorn även styra nätverkets massflödeshastighet. Resultaten visar att tra-ditionella reglerstrategier för fjärrvärmenätet kan förbättras med en MPC reg-ulator med framkoppling och att då även topplasten av värmeproduktionen minskar.

Konsumtionskraven från konsumenterna anslutna till fjärrvärmenätet kom-mer inte att förbli konstanta i framtiden. Värmepumpar är en livskraftig ut-manare till fjärrvärme, men de kan också komplettera ett lokalt värmenätverk i form av ett hybridvärmesystem. Den yttre påverkan som en värmepump ger när den körs parallellt med ett hem anslutet till fjärrvärme, ger insikt i hur fjär-rvärmeförbrukningsmönstret kan förändras över tid. Resultaten tyder på att en värmepump kan vara användbar i kombination med fjärrvärme under våren och hösten, vilket ger besparingar i uppvärmningskostnaderna samtidigt som den ger kyla under sommarmånaderna.

Nuvarande och framtida energisystem måste ses över holistiskt för att minska primär energiproduktion och -förbrukning, och lindra potentiell miljöförstöring. Utvecklingen av modelleringsbibliotek baserade på first-principle modelleringsteknik ger högre hållbarhet i energisystemsmod-elleringen. Trots att arbetet som presenteras i denna avhandling endast fokuserar på en typ av energiproduktion och förbrukning, så öppnar metoden i denna avhandling upp en flexibilitet som eliminerar behovet av entydig modellering och simulering genom att tillåta återanvändbarhet av modellkomponenter. Möjligheten att exportera dessa modeller särskiljer de ytterligare, eftersom de kan användas till att utvärdera nya reglerstrategier för energisystem när prediktioner i realtid blir mer tillgängliga.

v

I en simulering i sluten loop med börvärde på returtemperaturen kan MPC regulatorn även styra nätverkets massflödeshastighet. Resultaten visar att tra-ditionella reglerstrategier för fjärrvärmenätet kan förbättras med en MPC reg-ulator med framkoppling och att då även topplasten av värmeproduktionen minskar.

Konsumtionskraven från konsumenterna anslutna till fjärrvärmenätet kom-mer inte att förbli konstanta i framtiden. Värmepumpar är en livskraftig ut-manare till fjärrvärme, men de kan också komplettera ett lokalt värmenätverk i form av ett hybridvärmesystem. Den yttre påverkan som en värmepump ger när den körs parallellt med ett hem anslutet till fjärrvärme, ger insikt i hur fjär-rvärmeförbrukningsmönstret kan förändras över tid. Resultaten tyder på att en värmepump kan vara användbar i kombination med fjärrvärme under våren och hösten, vilket ger besparingar i uppvärmningskostnaderna samtidigt som den ger kyla under sommarmånaderna.

Nuvarande och framtida energisystem måste ses över holistiskt för att minska primär energiproduktion och -förbrukning, och lindra potentiell miljöförstöring. Utvecklingen av modelleringsbibliotek baserade på first-principle modelleringsteknik ger högre hållbarhet i energisystemsmod-elleringen. Trots att arbetet som presenteras i denna avhandling endast fokuserar på en typ av energiproduktion och förbrukning, så öppnar metoden i denna avhandling upp en flexibilitet som eliminerar behovet av entydig modellering och simulering genom att tillåta återanvändbarhet av modellkomponenter. Möjligheten att exportera dessa modeller särskiljer de ytterligare, eftersom de kan användas till att utvärdera nya reglerstrategier för energisystem när prediktioner i realtid blir mer tillgängliga.

(13)

Preface

This Ph.D. project has been carried out at the Division of Automation in Energy and Environmental Engineering at Mälardalen University. The course of the work spans from energy production to energy consumption in the heat and power sector. The modelling of waste fuel combustion began as a work package within the Polygeneration and Process Optimization for Advanced Combined Heat and Power Plants (POLYPO) Project under the Future Energy Research Profile at Mälardalen University along with the industrial partners Mälarenergi and Eskilstuna Energi och Miljö. My work and partnership with Mälarenergi continued to develop through the Future DIrections for Process industry Optimization (FUDIPO) project. During my time working with Mälarenergi, I developed an interest in district heating network behavior which was sparked by the start of the Smart Flow Project along with The Swedish Institute of Computer Science, ABB, Sigholm and EvoThings.

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

List of appended papers

I N. Zimmerman, K. Kyprianidis, C.F. Lindberg. (2016). Agglomeration Detection in Circulating Fluidized Bed Boilers using Refuse Derived Fuels. In proceedings of the 9th EUROSIM & the 57th SIMS. DOI: 10.3384/ecp17142148

II N. Zimmerman, K. Kyprianidis, C.F. Lindberg. (2018). Waste Fuel Combustion: Dynamic Modeling and Control. Processes, Special Issue Modeling and Simulation of Energy Systems, vol. 6, issue 11, DOI: 10.3390/pr6110222

III K. Hermansson, C. Kos, F. Starfelt, K. Kyprianidis, C.F. Lindberg, N. Zimmerman. (2018). An automated approach to building and simulating dynamic district heating networks. In proceedings of the International Federation of Automatic Control (IFAC), vol. 51-2, pp. 855-860, DOI: 10.1016/j.ifacol.2018.04.021

vi

Preface

This Ph.D. project has been carried out at the Division of Automation in Energy and Environmental Engineering at Mälardalen University. The course of the work spans from energy production to energy consumption in the heat and power sector. The modelling of waste fuel combustion began as a work package within the Polygeneration and Process Optimization for Advanced Combined Heat and Power Plants (POLYPO) Project under the Future Energy Research Profile at Mälardalen University along with the industrial partners Mälarenergi and Eskilstuna Energi och Miljö. My work and partnership with Mälarenergi continued to develop through the Future DIrections for Process industry Optimization (FUDIPO) project. During my time working with Mälarenergi, I developed an interest in district heating network behavior which was sparked by the start of the Smart Flow Project along with The Swedish Institute of Computer Science, ABB, Sigholm and EvoThings.

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

List of appended papers

I N. Zimmerman, K. Kyprianidis, C.F. Lindberg. (2016). Agglomeration Detection in Circulating Fluidized Bed Boilers using Refuse Derived Fuels. In proceedings of the 9th EUROSIM & the 57th SIMS. DOI: 10.3384/ecp17142148

II N. Zimmerman, K. Kyprianidis, C.F. Lindberg. (2018). Waste Fuel Combustion: Dynamic Modeling and Control. Processes, Special Issue Modeling and Simulation of Energy Systems, vol. 6, issue 11, DOI: 10.3390/pr6110222

III K. Hermansson, C. Kos, F. Starfelt, K. Kyprianidis, C.F. Lindberg, N. Zimmerman. (2018). An automated approach to building and simulating dynamic district heating networks. In proceedings of the International Federation of Automatic Control (IFAC), vol. 51-2, pp. 855-860, DOI: 10.1016/j.ifacol.2018.04.021

(14)

IV N. Zimmerman, K. Kyprianidis, C.F. Lindberg. (2019). Achieving Lower District Heating Network Temperatures Using Feed-Forward MPC. Materials, Special Issue on Advanced Control in the Energy Sector, vol. 12, issue 15, DOI: 10.3390/ma12152465

V F. Dall’Orto, N. Zimmerman, A. Vadiee, K. Kyprianidis. (2019). Eco-nomic Aspects of Hybrid Heating and Cooling Systems in a Residential Building. Submitted to the International conference on Applied Energy 2019.

*Reprints were made with permission from the publishers.

Contribution to papers

I was the main author and contributor in papers I, II, and IV, in which the co-authors contributed to manuscript conceptualization and subject-specific feedback and guidance. In papers III and V, I was responsible for drafting the manuscripts, which are based on the findings from my Masters Thesis stu-dents held at Mälardalen University. I was also involved in development of the problem statement, model development and methodology, and guidance. For all of the appended papers, I was also responsible for presenting the work and disseminating the findings at scientific peer-reviewed conferences, university seminars and project meetings.

Research articles not Included

The following publications and contributions by the author are not included in this thesis.

VI J. Campillo, N. Ghaviha, N. Zimmerman, E. Dahlquist. (2015). Flow batteries use potential in heavy vehicles. 2015 International conference on electrical systems for aircraft, railway, ship propulsion and road ve-hicles, IEEE, DOI: 10.1109/ESARS.2015.7101496

VII N. Zimmerman, E. Dahlquist, K. Kyprianidis. (2017). Towards On-line Fault Detection and Diagnostics in District Heating Systems. Energy Procedia 105, 1960-1966

VIII I. Aslanidou, V. Zaccaria, E. Pontika, N. Zimmerman, A. Kalfas, K. Kyprianidis. (2018). Teaching Gas Turbine Technology to Undergrad-uate Students in Sweden. ASME Turbo Expo 2018 best paper award, DOI: 10.1115/GT2018-77074

IX I. Aslanidou, N. Zimmerman, E. Pontika, A. Kalfas, K. Kyprianidis. (2020). Reforming heat and power technology course using student vii

IV N. Zimmerman, K. Kyprianidis, C.F. Lindberg. (2019). Achieving Lower District Heating Network Temperatures Using Feed-Forward MPC. Materials, Special Issue on Advanced Control in the Energy Sector, vol. 12, issue 15, DOI: 10.3390/ma12152465

V F. Dall’Orto, N. Zimmerman, A. Vadiee, K. Kyprianidis. (2019). Eco-nomic Aspects of Hybrid Heating and Cooling Systems in a Residential Building. Submitted to the International conference on Applied Energy 2019.

*Reprints were made with permission from the publishers.

Contribution to papers

I was the main author and contributor in papers I, II, and IV, in which the co-authors contributed to manuscript conceptualization and subject-specific feedback and guidance. In papers III and V, I was responsible for drafting the manuscripts, which are based on the findings from my Masters Thesis stu-dents held at Mälardalen University. I was also involved in development of the problem statement, model development and methodology, and guidance. For all of the appended papers, I was also responsible for presenting the work and disseminating the findings at scientific peer-reviewed conferences, university seminars and project meetings.

Research articles not Included

The following publications and contributions by the author are not included in this thesis.

VI J. Campillo, N. Ghaviha, N. Zimmerman, E. Dahlquist. (2015). Flow batteries use potential in heavy vehicles. 2015 International conference on electrical systems for aircraft, railway, ship propulsion and road ve-hicles, IEEE, DOI: 10.1109/ESARS.2015.7101496

VII N. Zimmerman, E. Dahlquist, K. Kyprianidis. (2017). Towards On-line Fault Detection and Diagnostics in District Heating Systems. Energy Procedia 105, 1960-1966

VIII I. Aslanidou, V. Zaccaria, E. Pontika, N. Zimmerman, A. Kalfas, K. Kyprianidis. (2018). Teaching Gas Turbine Technology to Undergrad-uate Students in Sweden. ASME Turbo Expo 2018 best paper award, DOI: 10.1115/GT2018-77074

IX I. Aslanidou, N. Zimmerman, E. Pontika, A. Kalfas, K. Kyprianidis. (2020). Reforming heat and power technology course using student vii

(15)

feedback to enhance learning experience. International Journal of Me-chanical Enineering Education, DOI: 10.1177/0306419019899939

viii

feedback to enhance learning experience. International Journal of Me-chanical Enineering Education, DOI: 10.1177/0306419019899939

(16)

List of Figures

Page

Figure 1.1: GHG emissions by sector in EU-28 for 2017. . . 2

Figure 1.2: Primary power production in EU-28 for 2017. . . 2

Figure 1.3: Objectives for the studies in the thesis. . . 3

Figure 1.4: Underlying connection of the presented research and

associated papers. . . 5

Figure 1.5: The percentage of bed material, within a size

distri-bution of 0.4 < dp < 0.63 mm, and correlation to the

mimimum fluidization velocity . . . 6

Figure 1.6: Measured output response to a step change in the

mea-sured disturbance. . . 7

Figure 1.7: Network heat propagation from 28 h to 34 h in the

aggregated model. . . 8

Figure 1.8: Excerpt of control results achieved by using the

his-toric load as the feed-forward signal. . . 9

Figure 1.9: Research methodology steps applied in this thesis. . . . 11 Figure 2.1: Circulating fluidized bed boiler. . . 13 Figure 2.2: District heating network concept. . . 17 Figure 2.3: Categorical breakdown of heat delivery in 2017 for

different parts of the world. . . 17 Figure 2.4: District heating network supply and return

tempera-tures as a function of outdoor temperature. . . 18 Figure 2.5: Seasonal variation in district heating network

temper-atures. . . 19 Figure 3.1: General system context. . . 24 Figure 3.2: Energy balance in a control volume. . . 26 Figure 3.3: Iterative approach towards model predictive control. . . 28 Figure 3.4: Example for reviewing controller performance. . . 31 Figure 3.5: The general concept of model predictive control. . . 32 Figure 4.1: Circulating fluidized bed boiler object-oriented

mod-elling approach in Dymola. . . 36 Figure 4.2: Dynamic model required inputs. . . 37 x

List of Figures

Page

Figure 1.1: GHG emissions by sector in EU-28 for 2017. . . 2

Figure 1.2: Primary power production in EU-28 for 2017. . . 2

Figure 1.3: Objectives for the studies in the thesis. . . 3

Figure 1.4: Underlying connection of the presented research and

associated papers. . . 5

Figure 1.5: The percentage of bed material, within a size

distri-bution of 0.4 < dp < 0.63 mm, and correlation to the

mimimum fluidization velocity . . . 6

Figure 1.6: Measured output response to a step change in the

mea-sured disturbance. . . 7

Figure 1.7: Network heat propagation from 28 h to 34 h in the

aggregated model. . . 8

Figure 1.8: Excerpt of control results achieved by using the

his-toric load as the feed-forward signal. . . 9

Figure 1.9: Research methodology steps applied in this thesis. . . . 11 Figure 2.1: Circulating fluidized bed boiler. . . 13 Figure 2.2: District heating network concept. . . 17 Figure 2.3: Categorical breakdown of heat delivery in 2017 for

different parts of the world. . . 17 Figure 2.4: District heating network supply and return

tempera-tures as a function of outdoor temperature. . . 18 Figure 2.5: Seasonal variation in district heating network

temper-atures. . . 19 Figure 3.1: General system context. . . 24 Figure 3.2: Energy balance in a control volume. . . 26 Figure 3.3: Iterative approach towards model predictive control. . . 28 Figure 3.4: Example for reviewing controller performance. . . 31 Figure 3.5: The general concept of model predictive control. . . 32 Figure 4.1: Circulating fluidized bed boiler object-oriented

mod-elling approach in Dymola. . . 36 Figure 4.2: Dynamic model required inputs. . . 37 x

(18)

Figure 4.3: Validation of CFB model: simulated predicted values in comparison to actual process signals. . . 38 Figure 4.4: Simplified representation of a district heating network

connected to six outlying regions with observed mass flow rate boundaries. . . 39 Figure 4.5: Validation results: simulated vs actual temperature

profiles supplied to the six outlying regions of Västerås. 40 Figure 4.6: Network delays as a function of the mass flow rate in

each identified distribution line. . . 41 Figure 4.7: Validation of DHN heat propagation to Tillberga . . . 42 Figure 4.8: Output response due to a step-change in the fuel’s

heating value comparing FFMPC, NFFMPC, and PI control schemes. . . 44 Figure 4.9: Control variable response due to a step-change in

the fuel’s heating value with a comparison between FFMPC, NFFMPC, and PI control schemes. . . 45 Figure 4.10: The results of implementing FFMPC to Tillberga. . . 47 Figure 4.11: Achievable improvements from FFMPC on a district

heating network. . . 48

xi

Figure 4.3: Validation of CFB model: simulated predicted values in comparison to actual process signals. . . 38 Figure 4.4: Simplified representation of a district heating network

connected to six outlying regions with observed mass flow rate boundaries. . . 39 Figure 4.5: Validation results: simulated vs actual temperature

profiles supplied to the six outlying regions of Västerås. 40 Figure 4.6: Network delays as a function of the mass flow rate in

each identified distribution line. . . 41 Figure 4.7: Validation of DHN heat propagation to Tillberga . . . 42 Figure 4.8: Output response due to a step-change in the fuel’s

heating value comparing FFMPC, NFFMPC, and PI control schemes. . . 44 Figure 4.9: Control variable response due to a step-change in

the fuel’s heating value with a comparison between FFMPC, NFFMPC, and PI control schemes. . . 45 Figure 4.10: The results of implementing FFMPC to Tillberga. . . 47 Figure 4.11: Achievable improvements from FFMPC on a district

heating network. . . 48

(19)

List of Tables

Page Table 1.1: Research questions and the papers, including answers to

research questions. . . 5

Table 2.1: Evolution of district heating. . . 18 Table 4.1: Modelling results for the six identified regions within the

DHN. . . 40

xii

List of Tables

Page Table 1.1: Research questions and the papers, including answers to

research questions. . . 5

Table 2.1: Evolution of district heating. . . 18 Table 4.1: Modelling results for the six identified regions within the

DHN. . . 40

(20)

Nomenclature

Abbreviations

ARX Auto Regressive with Exogenous input

CFB Circulating Fluidized Bed

CHP Combined Heat and Power

CV Control variable

DCS Distributed Control System

DH District Heating

DHN District Heating Network

DHW Domestic Hot Water

FFMPC Feed-Forward Model Predictive Control

GHG Greenhouse Gas

HEX Heat Exchanger

HP Heat pump

HV Heating Value

LHV Lower Heating Value

MD Measured Disturbance

MIMO Multiple-Input Multiple-Output

MO Measured Output

MPC Model Predictive Control

N4SID Numerical Subspace State-Space System Identification

NFFMPC No Feed-Forward Model Predictive Control

PI Proportional-Integral

PID Proportional-Integral-Derivative

xiii

Nomenclature

Abbreviations

ARX Auto Regressive with Exogenous input

CFB Circulating Fluidized Bed

CHP Combined Heat and Power

CV Control variable

DCS Distributed Control System

DH District Heating

DHN District Heating Network

DHW Domestic Hot Water

FFMPC Feed-Forward Model Predictive Control

GHG Greenhouse Gas

HEX Heat Exchanger

HP Heat pump

HV Heating Value

LHV Lower Heating Value

MD Measured Disturbance

MIMO Multiple-Input Multiple-Output

MO Measured Output

MPC Model Predictive Control

N4SID Numerical Subspace State-Space System Identification

NFFMPC No Feed-Forward Model Predictive Control

PI Proportional-Integral

PID Proportional-Integral-Derivative

(21)

RDF Refuse-Derived Fuel

RMSE Root Mean Square Error

SI System Identification

Greek Symbols

εmf Void fraction at minimum fluidization [-]

µg Gas viscosity [kgm−1s−1]

φ Sphericity [-]

ρ Density [kgm−3_]

ρg Gas density [kgm−3]

ρp Sand particle density [kgm−3]

τ Time delay [s]

Subscripts

i Control volume [-]

re f Reference value of quantity [-]

Latin Symbols

˙m Mass flow rate [kgs−1_]

˙

Q Heat [kW]

˙

W Work [kW]

ˆy Set-point of output(s) [-]

A Cross-sectional area [m2_]

c Control horizon [-]

cp Specific heat [kJkg−1_K−1_]

dp Sand particle diameter [µm]

g Gravitation constant ms−2_]

h Enthalpy [kJkg−1_]

k Current time step [-]

M Mass [kg]

p Prediction horizon [-]

xiv

RDF Refuse-Derived Fuel

RMSE Root Mean Square Error

SI System Identification

Greek Symbols

εmf Void fraction at minimum fluidization [-]

µg Gas viscosity [kgm−1s−1]

φ Sphericity [-]

ρ Density [kgm−3_]

ρg Gas density [kgm−3]

ρp Sand particle density [kgm−3]

τ Time delay [s]

Subscripts

i Control volume [-]

re f Reference value of quantity [-]

Latin Symbols

˙m Mass flow rate [kgs−1_]

˙

Q Heat [kW]

˙

W Work [kW]

ˆy Set-point of output(s) [-]

A Cross-sectional area [m2_]

c Control horizon [-]

cp Specific heat [kJkg−1_K−1_]

dp Sand particle diameter [µm]

g Gravitation constant ms−2_]

h Enthalpy [kJkg−1_]

k Current time step [-]

M Mass [kg]

p Prediction horizon [-]

(22)

Q Heat [kW]

Qi Controller output(s) weights [-]

r Pearson’s correlation coefficient [-]

Ri Weights of the control action [-]

T Temperature [K]

U Heat transfer coefficient [kWm−2_K−1_]

u Controller output(s) [-]

Um f Minimum fluidization [ms−1]

y Process output(s) [-]

xv

Q Heat [kW]

Qi Controller output(s) weights [-]

r Pearson’s correlation coefficient [-]

Ri Weights of the control action [-]

T Temperature [K]

U Heat transfer coefficient [kWm−2_K−1_]

u Controller output(s) [-]

Um f Minimum fluidization [ms−1]

y Process output(s) [-]

(23)

(24)

1. Introduction

This chapter presents a brief background motivating the work driving this thesis and the objectives to be achieved. Based on the objectives, the research questions are posed, followed by a summary of the appended articles. The challenges and assumptions made for the appended articles are addressed, and then the research approach utilized throughout the work is presented. The chapter concludes with an introduction of the thesis outline.

1.1 Background and Motivation

A growing challenge for the operational stability of thermal power plants is the growing influx of renewable solid fuel sources. The utilization of renew-able energy sources for the production of heat and power in combined heat and power plants (CHPs) is a driving force towards meeting environmental and societal goals. Popular research directives such as the Horizon 2020 ob-jectives are a means in the area of innovation for securing clean and efficient energy production within the European Union. It is important that mankind es-tablishes a good understanding of resource allocation and the environmental implications in order to derive an energy system that is secure, clean, efficient and sustainable (Z. Liu, 2015).

The implications of greenhouse gas (GHG) emissions for climate change are a global issue. It is imperative that GHG emissions are reduced. The heat and power generation sector is a leading contributor of greehouse gas sions(Zheng et al., 2019), accounting for more than 50% of the total emis-sions in the European Union’s 28 member countries (EU-28), as illustrated in Figure 1.1. This can be accomplished by reducing the consumption of fossil-based fuels by transitioning towards renewable fuels such as biomass and waste derivatives. Furthermore, the inefficient use of fuels can be resolved with better understanding of their roles in primary energy production and en-ergy consumption requirements. Both avenues can lead to a more efficient system and reduce GHG emissions. Figure 1.2 outlines the breakdown of pri-mary energy production in 2017 for the EU-28. Fossil-based fuels account for 38.82% of total primary energy production, renewable energy sources repre-sent 29.88% and bioenergy and waste account for 18.09%. A more efficient means to primary power production is through CHPs, which can produce both heat and power simultaneously. A shift towards more efficient systems is

es-1

1. Introduction

This chapter presents a brief background motivating the work driving this thesis and the objectives to be achieved. Based on the objectives, the research questions are posed, followed by a summary of the appended articles. The challenges and assumptions made for the appended articles are addressed, and then the research approach utilized throughout the work is presented. The chapter concludes with an introduction of the thesis outline.

1.1 Background and Motivation

A growing challenge for the operational stability of thermal power plants is the growing influx of renewable solid fuel sources. The utilization of renew-able energy sources for the production of heat and power in combined heat and power plants (CHPs) is a driving force towards meeting environmental and societal goals. Popular research directives such as the Horizon 2020 ob-jectives are a means in the area of innovation for securing clean and efficient energy production within the European Union. It is important that mankind es-tablishes a good understanding of resource allocation and the environmental implications in order to derive an energy system that is secure, clean, efficient and sustainable (Z. Liu, 2015).

The implications of greenhouse gas (GHG) emissions for climate change are a global issue. It is imperative that GHG emissions are reduced. The heat and power generation sector is a leading contributor of greehouse gas sions(Zheng et al., 2019), accounting for more than 50% of the total emis-sions in the European Union’s 28 member countries (EU-28), as illustrated in Figure 1.1. This can be accomplished by reducing the consumption of fossil-based fuels by transitioning towards renewable fuels such as biomass and waste derivatives. Furthermore, the inefficient use of fuels can be resolved with better understanding of their roles in primary energy production and en-ergy consumption requirements. Both avenues can lead to a more efficient system and reduce GHG emissions. Figure 1.2 outlines the breakdown of pri-mary energy production in 2017 for the EU-28. Fossil-based fuels account for 38.82% of total primary energy production, renewable energy sources repre-sent 29.88% and bioenergy and waste account for 18.09%. A more efficient means to primary power production is through CHPs, which can produce both heat and power simultaneously. A shift towards more efficient systems is

(25)

2

Figure 1.1: GHG emissions by sector in the EU-28 for 2017, reproduced from (Euro-stat, 2019b).

Figure 1.2: Primary power production in EU-28 for 2017, reproduced from (Eurostat, 2019a) .

sential because it has been estimated that by 2040 the demand for energy will grow by more than 25% (International Energy Agency, 2018). Therefore, an investment in CHP plants will help to mitigate the future increase of en-ergy requirements and the associated GHG emissions. Circulating fluidized bed boilers (CFBs) are one type of technology which can be implemented for

CHP production because they can:(i) use a range of fuels with diverse

compo-sitions, e.g. biomass and waste,(ii) achieve high combustion efficiency, (iii)

result in lower emissions, and(iv) adjust quickly to load changes (Gungor,

2009; Van Caneghem et al., 2012; Basu & Fraser, 2015).

There are many areas from which energy savings and emissions reductions can be achieved in CHPs. However, the following two key areas are the focus of this work:

2

Figure 1.1: GHG emissions by sector in the EU-28 for 2017, reproduced from (Euro-stat, 2019b).

Figure 1.2: Primary power production in EU-28 for 2017, reproduced from (Eurostat, 2019a) .

sential because it has been estimated that by 2040 the demand for energy will grow by more than 25% (International Energy Agency, 2018). Therefore, an investment in CHP plants will help to mitigate the future increase of en-ergy requirements and the associated GHG emissions. Circulating fluidized bed boilers (CFBs) are one type of technology which can be implemented for

CHP production because they can:(i) use a range of fuels with diverse

compo-sitions, e.g. biomass and waste,(ii) achieve high combustion efficiency, (iii)

result in lower emissions, and(iv) adjust quickly to load changes (Gungor,

2009; Van Caneghem et al., 2012; Basu & Fraser, 2015).

There are many areas from which energy savings and emissions reductions can be achieved in CHPs. However, the following two key areas are the focus of this work:

(26)

1 Thermal treatment of refuse-derived fuels in CFBs. 2 Heat propagation in district heating networks.

The objectives of this work to identify the areas from which process improve-ments can be obtained are illustrated in Figure 1.3, which relates to the papers presented in section 1.3. Models have been developed to further understand the transient behaviors associated with the combustion of refuse-derived fuel (RDF) in CFBs and the demand for district heating (DH) within district heat-ing networks (DHNs). Control approaches have been derived for both domains and it was found that there is potential for a higher level of operational perfor-mance and control strategies. The evaluation of key proponents contributing to process disturbances, e.g. transient behavior, leads to the ability to improve process performance by introducing feed-forward model predictive control (FFMPC) for improved disturbance rejection. The added impact of introduc-ing feed-forward signal(s) affords the ability to design a controller that can work proactively when information on process disturbances is available. The five papers represented in Figure 1.3 are categorized by their applicability to the current work in section 1.3.

Figure 1.3: Objectives for the studies in the thesis (CFB - circulating fluidized bed boiler, DHN - district heating network).

1.2 Research Questions

Based on the defined objectives, two areas of research are considered and expanded upon in response to two subsequent research questions. The first research area relates to the development of dynamic models. The following research questions are formulated based on the literature analysis:

1 Thermal treatment of refuse-derived fuels in CFBs. 2 Heat propagation in district heating networks.

The objectives of this work to identify the areas from which process improve-ments can be obtained are illustrated in Figure 1.3, which relates to the papers presented in section 1.3. Models have been developed to further understand the transient behaviors associated with the combustion of refuse-derived fuel (RDF) in CFBs and the demand for district heating (DH) within district heat-ing networks (DHNs). Control approaches have been derived for both domains and it was found that there is potential for a higher level of operational perfor-mance and control strategies. The evaluation of key proponents contributing to process disturbances, e.g. transient behavior, leads to the ability to improve process performance by introducing feed-forward model predictive control (FFMPC) for improved disturbance rejection. The added impact of introduc-ing feed-forward signal(s) affords the ability to design a controller that can work proactively when information on process disturbances is available. The five papers represented in Figure 1.3 are categorized by their applicability to the current work in section 1.3.

Figure 1.3: Objectives for the studies in the thesis (CFB - circulating fluidized bed boiler, DHN - district heating network).

1.2 Research Questions

Based on the defined objectives, two areas of research are considered and expanded upon in response to two subsequent research questions. The first research area relates to the development of dynamic models. The following research questions are formulated based on the literature analysis:

(27)

4

RQ1: What are the challenges associated with modelling the combustion of waste fuel, and what level of fidelity is permissible for derived models to predict combustion temperatures under transient behav-ior in order to be applicable in different control schemes?

RQ2: What is the degree of fidelity required for modelling thermal heat-ing networks to assist in system performance when external forces are at play, and how should the derived models be refined in order to test a control strategy?

The second research area relates to controlling the dynamic models developed within the first research area. Based on the literature analysis, the following research questions are formulated:

RQ3: What are the potential and limitations of model predictive con-trol for improving process performance in RDF-fired CFBs in the presence of feed-forward information?

RQ4: What is the potential for implementing FFMPC in district heating networks in order to achieve higher system performance in the presence of real-time predictions?

1.3 Summary of Included Articles

This section begins with a brief illustration of how the concepts of modelling and control are connected to the domains of the appended articles. The articles are also organized in relation to the presented research questions. The section concludes with a summary of the appended papers, highlighting the key findings and their relationships to the research questions.

This doctoral thesis comprises the five appended peer-reviewed papers. The main topics of this thesis lie within the domains of modelling and control of CFBs and DHNs, illustrated in Figure 1.4. The answers to each research question are presented in Table 1.1.

4

RQ1: What are the challenges associated with modelling the combustion of waste fuel, and what level of fidelity is permissible for derived models to predict combustion temperatures under transient behav-ior in order to be applicable in different control schemes?

RQ2: What is the degree of fidelity required for modelling thermal heat-ing networks to assist in system performance when external forces are at play, and how should the derived models be refined in order to test a control strategy?

The second research area relates to controlling the dynamic models developed within the first research area. Based on the literature analysis, the following research questions are formulated:

RQ3: What are the potential and limitations of model predictive con-trol for improving process performance in RDF-fired CFBs in the presence of feed-forward information?

RQ4: What is the potential for implementing FFMPC in district heating networks in order to achieve higher system performance in the presence of real-time predictions?

1.3 Summary of Included Articles

This section begins with a brief illustration of how the concepts of modelling and control are connected to the domains of the appended articles. The articles are also organized in relation to the presented research questions. The section concludes with a summary of the appended papers, highlighting the key findings and their relationships to the research questions.

This doctoral thesis comprises the five appended peer-reviewed papers. The main topics of this thesis lie within the domains of modelling and control of CFBs and DHNs, illustrated in Figure 1.4. The answers to each research question are presented in Table 1.1.

(28)

Figure 1.4: Underlying connection of the presented research and associated papers. Table 1.1: Research questions and the papers,

in-cluding answers to research questions.

Research Question Article

RQ 1 Paper I & II

RQ 2 Paper III, IV & V

RQ 3 Paper II

RQ 4 Paper IV

Paper I: Agglomeration Detection in Circulating Fluidized Bed

Boilers using Refuse Derived Fuels

The work presented in paper I presents the methodology for modelling

the transient combustion behavior of refuse derived fuel. The presented

methodology is a step towards answering RQ1, which is associated with

the complications of modelling such a process. A focal point of the paper is supplementing the derived physics-based model with the ability to estimate the minimum fluidization velocity required within the boiler. The underlying theory is that by predicting the minimum fluidization it will be possible to predict bed-material agglomeration. The composition of RDF is highly heterogeneous, but the composition and size distribution of sand is markedly

Figure 1.4: Underlying connection of the presented research and associated papers. Table 1.1: Research questions and the papers,

in-cluding answers to research questions.

Research Question Article

RQ 1 Paper I & II

RQ 2 Paper III, IV & V

RQ 3 Paper II

RQ 4 Paper IV

Paper I: Agglomeration Detection in Circulating Fluidized Bed

Boilers using Refuse Derived Fuels

The work presented in paper I presents the methodology for modelling

the transient combustion behavior of refuse derived fuel. The presented

methodology is a step towards answering RQ1, which is associated with

the complications of modelling such a process. A focal point of the paper is supplementing the derived physics-based model with the ability to estimate the minimum fluidization velocity required within the boiler. The underlying theory is that by predicting the minimum fluidization it will be possible to predict bed-material agglomeration. The composition of RDF is highly heterogeneous, but the composition and size distribution of sand is markedly

(29)

6

more consistent. With an average diameter in the range of 0.2 < dp <0.4

the facilitating affects of agglomerates, i.e. the buildup of material around a sand particle, will necessitate a higher minimum fluidization velocity as the average weight of the bed material increases above that of sand alone. Figure 1.5 illustrates the supporting results, which validate this proposition. The figure shows that the bed material size distribution trends along with the minimum fluidization velocity over a testing period.

1 2 3 4 5 6 7 Time [d] 10 15 20 25 30 35 40 Size Distribution [%] 0.18 0.185 0.19 0.195 0.2 0.205 0.21 Minimum Fluidization [m s -1] 0.4-0.63 mm umf

Figure 1.5: The percentage of bed material, within a size distribution of 0.4 < dp<

0.63 mm, and correlation to the mimimum fluidization velocity

Paper II: Waste Fuel Combustion: Dynamic Modelling and

Control

The work presented inpaper II is directed at answering RQ1 and RQ3. The

work is a continuation of the physical models derived inpaper I. Based on

the validated physical model, the feasibility of implementing model predictive control (MPC) is outlined and analysed. Two different MPCs are introduced: FFMPC and MPC without feed-forward (NFFMPC). The MPCs are then com-pared with the more traditional approach of proportional-integral (PI)-based control. The validated physical model allows for the identification of a dis-crete linear state-space model. This, in turn, is implemented as the MPC’s internal model based on the assumption that the physical model can predict the boiler bed and riser temperatures, live steam temperature and boiler load. The feed-forward aspect in the approach for FFMPC is accomplished by intro-ducing the fuel’s heating value as if it is a known measured disturbance, and therefore grants the controller the ability to proactively adjust the manipulable variables in the process. In the NFFMPC approach, the controller’s response

6

more consistent. With an average diameter in the range of 0.2 < dp < 0.4

the facilitating affects of agglomerates, i.e. the buildup of material around a sand particle, will necessitate a higher minimum fluidization velocity as the average weight of the bed material increases above that of sand alone. Figure 1.5 illustrates the supporting results, which validate this proposition. The figure shows that the bed material size distribution trends along with the minimum fluidization velocity over a testing period.

1 2 3 4 5 6 7 Time [d] 10 15 20 25 30 35 40 Size Distribution [%] 0.18 0.185 0.19 0.195 0.2 0.205 0.21 Minimum Fluidization [m s -1] 0.4-0.63 mm umf

Figure 1.5: The percentage of bed material, within a size distribution of 0.4 < dp<

0.63 mm, and correlation to the mimimum fluidization velocity

Paper II: Waste Fuel Combustion: Dynamic Modelling and

Control

The work presented inpaper II is directed at answering RQ1 and RQ3. The

work is a continuation of the physical models derived inpaper I. Based on

the validated physical model, the feasibility of implementing model predictive control (MPC) is outlined and analysed. Two different MPCs are introduced: FFMPC and MPC without feed-forward (NFFMPC). The MPCs are then com-pared with the more traditional approach of proportional-integral (PI)-based control. The validated physical model allows for the identification of a dis-crete linear state-space model. This, in turn, is implemented as the MPC’s internal model based on the assumption that the physical model can predict the boiler bed and riser temperatures, live steam temperature and boiler load. The feed-forward aspect in the approach for FFMPC is accomplished by intro-ducing the fuel’s heating value as if it is a known measured disturbance, and therefore grants the controller the ability to proactively adjust the manipulable variables in the process. In the NFFMPC approach, the controller’s response