Ng, Lik Yin (2015) Novel integrated design techniques for biorefineries. PhD thesis, University of Nottingham.

Transcription

1 Ng, Lik Yin (2015) Novel integrated design techniques for biorefineries. PhD thesis, University of Nottingham. Access from the University of Nottingham reository: htt://erints.nottingham.ac.uk/29016/1/thesis_ng%20lik%20yin.df Coyright and reuse: The Nottingham eprints service makes this work by researchers of the University of Nottingham available oen access under the following conditions. This article is made available under the University of Nottingham End User licence and may be reused according to the conditions of the licence. For more details see: htt://erints.nottingham.ac.uk/end_user_agreement.df For more information, lease contact

2 NOVEL INTEGRATED DESIGN TECHNIQUES FOR BIOREFINERIES LIK YIN NG, MEng. Thesis submitted to the University of Nottingham for the degree of Doctor of Philosohy JULY 2015

3 Abstract ABSTRACT Utilisation of biomass is identified as one of the romising solutions to reduce society s deendence on fossil fuels and mitigate climate change caused by the exloitation of fossil fuels. By using the concet of biorefinery, biomass can be converted into value-added roducts such as biofuels, biochemical roducts and biomaterials in a greener and sustainable way. To enhance the efficiency of biorefinery, the concet of integrated biorefinery which focuses on the integration of various biomass conversion technologies is utilised. To date, various biomass conversion athways are available to convert biomass into a wide range of roducts. Due to the substantial amount of otential roducts and conversion technologies, determining of chemical roducts and rocessing routes in an integrated biorefinery have become more challenging. Hence, there is a need for a methodology caable of evaluating the integrated rocess in order to identify the otimal roducts as well as the otimal conversion athways that roduce the identified roducts. This thesis resents a novel aroach which integrates rocess with roduct design techniques for integrated biorefineries. In the roosed aroach, integration between synthesis of integrated biorefinery and comuter-aided molecular design (CAMD) techniques is resented. By using CAMD techniques, otimal chemical roduct in terms of target roerties i

4 Abstract which fulfils the required roduct needs is designed. On the other hand, in order to identify the conversion athways that roduce the identified otimal chemical roduct in an integrated biorefinery, chemical reaction athway ma (CRPM) and suerstructural mathematical otimisation aroach have been utilised. Furthermore, this thesis also resents various chemical roduct design aroaches. In order to solve chemical design roblems where multile roduct needs are required to be considered and otimised, a novel multi-objective otimisation aroach for chemical roduct design has been resented. By using fuzzy otimisation aroach, the develoed multiobjective otimisation aroach identifies otimal chemical roduct based on multile roduct roerties. In addition, fuzzy otimisation aroach has been further extended to address chemical roduct design roblems where the accuracy of roerty rediction model is taken into account. A robust chemical roduct design aroach is develoed to design otimal chemical roducts with consideration of accuracy of roerty rediction model. Furthermore, together with CAMD techniques and suerstructural mathematical otimisation aroach, the develoed multi-objective otimisation aroach has been utilised for the design of mixtures in an integrated biorefinery. For this urose, a systematic otimisation aroach has been develoed to identify otimal mixture based on multile desired roduct needs as well as the otimal conversion athways that convert biomass into the otimal mixture. Finally, ossible extensions and future oortunities for the realm of the research work have been highlighted in the later art of this thesis. ii

5 List of Publications LIST OF PUBLICATIONS Refereed journals Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2014). A multiobjective otimization based aroach for otimal chemical roduct design, Industrial & Engineering Chemistry Research, 53, 44, , DOI: /ie502906a Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2014). Robust chemical roduct design via fuzzy otimisation aroach, Comuters & Chemical Engineering, DOI: /j.comchemeng (in ress) Ng, L. Y., Chong, F. K., & Chemmangattuvalail, N. G. (2014). Challenges and oortunities in comuter-aided molecular design, Comuters & Chemical Engineering, DOI: /j.comchemeng (in ress) Ng, L. Y., Andiaan, V., Chemmangattuvalail, N. G., & Ng, D. K. S. (2014). A systematic methodology for otimal mixture design in an integrated biorefinery, Comuters & Chemical Engineering, DOI: /j.comchemeng (in ress) iii

6 List of Publications Ng, L. Y., Andiaan, V., Chemmangattuvalail, N. G., & Ng, D. K. S. (2015). Novel methodology for the synthesis of otimal biochemicals in integrated biorefineries via inverse design techniques, Industrial & Engineering Chemistry Research, DOI: /acs.iecr.5b00217 (in ress) Andiaan, V., Ko, A. S. Y., Lau, V. W. S., Ng, L. Y., Ng, R. T. L., Chemmangattuvalail, N. G., & Ng, D. K. S. (2015). Synthesis of sustainable integrated biorefinery via reaction athway synthesis: Economic, incremental environmental burden and energy assessment with multiobjective otimization, AIChE Journal, 61, 1, , DOI: /aic Book chater Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2015). Novel methodologies for otimal roduct design from biomass, in Ng, D. K. S., Tan, R. R., El-Halwagi, M. M. & Foo, D. C. Y. (Eds), Process Design Strategies for Biomass Conversion Systems, Wiley & Sons, Inc. (acceted) Keynote lecture Ng, L. Y., Chong, F. K., & Chemmangattuvalail, N. G. (2014). Challenges and oortunities in comuter aided molecular design, 8 th International Conference on Foundations of Comuter-Aided Process Design, July 2014, Cle Elum, Washington, United States, Paer no.: 129. iv

7 List of Publications Conference resentations Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2013). Synthesis of otimal biofuels from alm based biomass via inverse design techniques. 6 th International Conference on Process Systems Engineering ASIA 2013, June 2013, Kuala Lumur, Malaysia, Paer no.: 33. Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2013). Design of otimal biochemical from alm based biomass via integrated biorefineriesinverse design techniques. Postgraduate Colloquium for Environmental Research 2013, June 2013, Genting Highlands, Malaysia, Paer no.: 36. Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2013). Otimal chemical roduct design via fuzzy otimisation based inverse design techniques. 9 th World Congress of Chemical Engineering Incororating 15 th Asian Pacific Confederation of Chemical Engineering Congress, August 2013, Seoul, Korea, Paer no.: O Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2014). Otimal chemical roduct design via fuzzy otimisation based inverse design techniques. 24 th Euroean Symosium on Comuter Aided Process Engineering, June 2014, Budaest, Hungary, Paer no.: 57. v

8 List of Publications Andiaan, V., Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2014). Systematic chemical reaction athway synthesis for sustainable integrated biorefineries. 8 th International Conference on Foundations of Comuter-Aided Process Design, July 2014, Cle Elum, Washington, United States, Paer no.: 12. Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2014). Robust chemical roduct design via fuzzy otimisation aroach. 8 th International Conference on Foundations of Comuter-Aided Process Design, July 2014, Cle Elum, Washington, United States, Paer no.: 14. Ng, L. Y., Chemmangattuvalail, N. G., & Ng, D. K. S. (2014). Robust chemical roduct design via fuzzy otimisation aroach. 27 th Symosium of Malaysian Chemical Engineers in Conjunction with 21 st Regional Symosium on Chemical Engineering, October 2014, Taylor s University, Lakeside Camus, Subang Jaya, Malaysia, Paer no.: PSE 060. Ng, L. Y., Andiaan, V., Chemmangattuvalail, N. G., & Ng, D. K. S. (2015). A systematic methodology for otimal mixture design in an integrated biorefinery. 12 th International Symosium on Process Systems Engineering and 25 th Euroean Symosium on Comuter Aided Process Engineering, 31 May-4 June 2015, Coenhagen, Denmark, Paer no.: 411. vi

9 Acknowledgement ACKNOWLEDGEMENT First and foremost, I would like to exress my greatest gratitude to my suervisors Dr. Nishanth G. Chemmangattuvalail and Dr. Denny K. S. Ng for their knowledge and guidance. Their constant suort, understanding as well as atience given to me are greatly areciated. Furthermore, I would like to thank Prof. Dominic C. Y. Foo for the advice and encouragement he has rovided. Thanks are also due to my friends and co-workers from Sustainable Process Integration research grou for the discussion and sharing of research ideas. I am also grateful to DDr. Hon Loong Lam for his effort and time in reviewing my annual reorts. In addition, I would like to extend my sincere thanks and areciation to the administrative staffs of the Faculty of Engineering, The University of Nottingham Malaysia Camus for their assistance and suort. I would also like to thank The University of Nottingham Malaysia Camus for their financial suort granted through Dean s PhD Scholarshi. A secial and heartfelt gratitude is given to my arents, my brother, sister-in-law and my sister for their endless love and suort. Last but not least, multile thanks to God, who made all things ossible. vii

10 Table of Contents TABLE OF CONTENTS ABSTRACT LIST OF PUBLICATIONS ACKNOWLEDGEMENT TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES LIST OF NOMENCLATURE i iii vii viii xiii xvii xx CHAPTER 1 INTRODUCTION Problem Background Integrated Biorefinery Chemical Product Design Problem Statement Research Objectives Outline of Thesis 14 CHAPTER 2 LITERATURE REVIEW Potential Products from Biomass Synthesis and Design of Integrated Biorefinery Comuter-Aided Molecular Design Tyes of Proerties and Estimation Techniques Grou Contribution Methods 38 viii

11 Table of Contents Toological Indices and Grou Contribution + Method Molecular Signature Descritor Current Status in Comuter-Aided Molecular Design Integrated Process and Molecular Design Mixture Design Multi-objective Chemical Product Design Robust Chemical Product Design Conclusions 72 CHAPTER 3 RESEARCH SCOPES AND METHODOLOGIES Research Scoes Research Methodology Concetual Design for Synthesis of Chemical Product from Biomass in Integrated Biorefineries Systematic Methodology for Otimal Chemical Product Design in Integrated Biorefineries Multi-objective Otimisation Aroach for Otimal Chemical Product Design Robust Chemical Product Design via Multi-objective Otimisation Aroach Systematic Methodology for Otimal Mixture Design in Integrated Biorefineries Conclusions 85 ix

12 Table of Contents CHAPTER 4 CONCEPTUAL DESIGN FOR SYNTHESIS OF CHEMICAL PRODUCT FROM BIOMASS IN INTEGRATED BIOREFINERIES Introduction Integration of Molecular Design Techniques and Chemical Reaction Pathway Synthesis Design of Biochemical Product: Signature Based Molecular Design Technique Design of Integrated Biorefinery: Chemical Reaction Pathway Synthesis Case Study Identification of Bio-based Fuel Identification of Chemical Reaction Pathway Conclusion 113 CHAPTER 5 SYSTEMATIC METHODOLOGY FOR OPTIMAL CHEMICAL PRODUCT DESIGN IN INTEGRATED BIOREFINERIES Introduction Two-stage Otimisation Aroach for Synthesis of Otimal Biochemical Products Stage 1: Design of Otimal Biochemical Product Stage 2: Design of Integrated Biorefinery Case Study Design of Otimal Product Selection of Otimal Conversion Pathway Conclusion 140 x

13 Table of Contents CHAPTER 6 MULTI-OBJECTIVE OPTIMISATION APPROACH FOR OPTIMAL CHEMICAL PRODUCT DESIGN Introduction Systematic Multi-objective Otimisation Aroach Fuzzy Otimisation Aroach Bi-level Otimisation Aroach Aroaches in Fuzzy Otimisation Solution Procedure Case Study Problem Descrition and Problem Statement Solution of Design Problem Results and Discussions Conclusion 173 CHAPTER 7 ROBUST CHEMICAL PRODUCT DESIGN VIA FUZZY OPTIMISATION APPROACH Introduction Robust Chemical Product Design Aroach Fuzzy Otimisation based Molecular Design Techniques under Uncertainty of Proerty Prediction Models Solution Procedure Case Study: Otimal Design of Alkyl Substituent for Fungicide Problem Descrition Problem Formulation Results and Discussions Conclusion 208 xi

14 Table of Contents CHAPTER 8 SYSTEMATIC METHODOLOGY FOR OPTIMAL MIXTURE DESIGN IN INTEGRATED BIOREFINERIES Introduction Two-stage Otimisation Aroach for Otimal Mixture Design Stage 1: Mixture Design Stage 2: Integrated biorefinery design Case Study: Mixture design for bio-based fuel additives Design of Otimal Mixture Selection of Otimal Conversion Pathway Conclusion 250 CHAPTER 9 CONCLUSIONS AND FUTURE WORK Conclusions Future Work Enhancement of Process and Product Details in Integrated Biorefineries Consideration of Business Asects in Integrated Designs Sustainable Design Framework Mixture Design 257 REFERENCES 259 APPENDICES 286 xii

15 List of Tables LIST OF TABLES Table 1.1: Classification of chemical roducts 6 Table 2.1: Classification of chemical roduct roerties 36 Table 4.1: List of athways and secifications for the develoed CRPM 101 Table 4.2: Uer and lower bounds for bio-based fuel design 104 Table 4.3: List of signatures for bio-based fuel design 106 Table 4.4: Normalised roerty oerators and normalised target roerty ranges for the bio-based fuel design roblem 107 Table 4.5: Bio-based fuel design solutions in terms of signatures 109 Table 4.6: Table 4.7: Table 5.1: Bio-based fuel design solutions in terms of roduct secifications 109 Bio-based fuel design solutions in terms of molecular structures 110 Uer and lower limits for bio-based fuel design roblem 123 Table 5.2: List of signatures 126 Table 5.3: Normalised roerty oerators and normalised target roerty ranges for the bio-based fuel design roblem 126 Table 5.4: List of solutions in terms of signatures 127 Table 5.5: List of solutions in terms of roduct secifications 128 xiii

16 List of Tables Table 5.6: List of solutions in terms of molecular structure 129 Table 5.7: Lignocellulosic comosition of EFB 130 Table 5.8: List of conversions and selectivities for conversion athways 131 Table 5.9: List of market rices of roducts and raw material 134 Table 5.10: List of caital and oerating costs for conversion athways 135 Table 5.11: Comarison of results for scenario 1 and Table 6.1: Table 6.2: Incomlete target roerty range and normalised target roerty ranges for the solvent design roblem 159 Comlete target roerty ranges and normalised target roerty ranges for the solvent design roblem 160 Table 6.3: List of solutions in terms of signatures 164 Table 6.4: List of solutions in terms of roduct secifications 165 Table 6.5: List of solutions in terms of molecular structure 166 Table 6.6: List of solutions with single otimised target roerty 167 Table 6.7: Comarison of λ between different solutions 168 Table 6.8: Comarison of * between different solutions 171 Table 7.1: Table 7.2: Lower and uer limits for regions with different certainty 181 Target roerty ranges and shifted target roerty ranges for the fungicide design roblem 192 Table 7.3: Fuzzy membershi functions of target roerties 193 xiv

17 List of Tables Table 7.4: List of generated height two signatures 194 Table 7.5: Table 7.6: Table 7.7: Table 7.8: Table 7.9: Table 7.10: List of solutions of alkyl substituents for DD in terms of signatures 198 List of solutions by using max-min aggregation aroach 198 Otimal solutions obtained by works from other authors 199 Designs of alkyl substituents for DD for different scenarios 201 List of solutions by using by using two hase aroach 205 Molecular structures for the solutions of fungicide design roblem 207 Table 8.1: Target roerty ranges for the mixture design 226 Table 8.2: Comonents and comosition of the main comonent 226 Table 8.3: Table 8.4: Target roerty ranges and target oerator ranges for the design of additive made from alkane 230 Target roerty ranges and target oerator ranges for the design of additive made from alcohol 231 Table 8.5: List of solutions of additive made from alkane 231 Table 8.6: List of solutions of additive made from alcohol 232 Table 8.7: Lower and uer limits for the mixing of target roerties 234 Table 8.8: List of solutions of fuel mixture 236 xv

18 List of Tables Table 8.9: Table 8.10: Comarison of λ between different designs of mixture 236 Hildebrand solubility arameters determined for main and additive comonents 238 Table 8.11: Lignocellulosic comosition of EFB 239 Table 8.12: List of conversions and selectivities for conversion athways 240 Table 8.13: List of rices of roducts and raw material 243 Table 8.14: List of caital and oerating costs for conversion athways 244 Table 8.15: Comarison of results for scenario 1 and Table S1: List of signature and molecular grous for 2,2,4- trimethylentane 287 Table S2: Calculation of H v for 2,2,4-trimethylentane 288 Table S3: Calculation for LC 50 for 2,2,4-trimethylentane 289 Table S4: List of signature and molecular grous for 2-entanol 290 Table S5: Calculation of H v for 2-entanol 291 Table S6: Calculation for LC 50 for 2-entanol 292 xvi

19 List of Figures LIST OF FIGURES Figure 2.1: Traditional molecular design aroach 30 Figure 2.2: Basic stes in generate-and-search CAMD aroach 32 Figure 2.3: Classification for roerty estimation methods 38 Figure 2.4: Proerty rediction by using QSPR model 42 Figure 2.5: Illustration of atomic signatures u to height 3 46 Figure 2.6: Molecular grah colouring of atomic signature 47 Figure 2.7: Figure 2.8: Figure 2.9: Exression of higher order molecular grous with signatures 48 Integrated rocess-roduct design for reverse roblem formulations 55 Fuzzy membershi functions for: (a) maximising; (b) minimising; (c) traezoidal-shaed; (d) triangular constraints 65 Figure 3.1: Research methodology 80 Figure 4.1: Figure 4.2: Integration of integrated biorefinery and molecular roduct design 88 Procedure for solving a chemical roduct design roblem 90 Figure 4.3: Exlanation of handshaking dilemma 96 Figure 4.4: Enumeration of molecular structure for roan-1-ol 98 xvii

20 List of Figures Figure 4.5: Biomass chemical reaction athway ma 100 Figure 4.6: Figure 5.1: Chemical reaction athway for the conversion of alm-based biomass into bio-based fuel 112 Suerstructure as reresentation of integrated biorefinery 117 Figure 5.2: General reresentation of integrated biorefinery 133 Figure 5.3: Figure 5.4: Figure 6.1: Figure 6.2: Flow diagram of synthesised integrated biorefinery (maximum roduct yield) 136 Flow diagram of synthesised integrated biorefinery (maximum economic otential) 138 Fuzzy degree of satisfaction (λ) of the inequalities: (a) roerty to be minimised; (b) roerty to be maximised 144 Procedure for solving a multi-objective chemical roduct design roblem 151 Figure 6.3: Process flow sheet of gas sweetening rocess 153 Figure 6.4: Pareto frontier for the solvent design roblem 172 Figure 7.1: Figure 7.2: Fuzzy membershi functions for: (a) roerty robustness; (b) roerty sueriority of roerty to be maximised; (c) roerty sueriority of roerty to be minimised 180 Procedure for solving a chemical roduct design roblem with consideration of roerty rediction model uncertainty 187 Figure 7.3: Chemical structure of DD 190 xviii

21 List of Figures Figure 7.4: Figure 8.1: Comarison of λ for the solutions of fungicide design roblem 203 Two-stage otimisation aroach to roduce otimal mixture from biomass 211 Figure 8.2: Procedure for solving a mixture design roblem 214 Figure 8.3: Figure 8.4: Figure 8.5: Production of additives made from alkane and alcohol from lignocellulosic biomass 242 Flow diagram of the synthesised integrated biorefinery (maximum roduct yield) 247 Flow diagram of the synthesised integrated biorefinery (maximum economic otential) 249 xix

22 List of Nomenclature LIST OF NOMENCLATURE Abbreviations AAE Average Absolute Error ARE Average Relative Error BB Branch and Bound CAFD Comuter-Aided Flow Sheet Design CAMD Comuter-Aided Molecular Design CAMPD Comuter-Aided Molecular and Process Design CoMT Continuous-Molecular Targeting CRPM Chemical Reaction Pathway Ma CPU Central Processing Unit CR Certain Region EFB Emty Fruit Bunch EGA Efficient Genetic Algorithm FT Fischer-Trosch GA Genetic Algorithm GC Grou Contribution xx

23 List of Nomenclature GHG Greenhouse Gas GWP Global Warming Potential HEN Heat Exchanger Network HMF Hydroxymethylfurfural HSGA Hammersley Stochastic Genetic Algorithm HSTA Hammersley Stochastic Annealing Algorithm IL Ionic Liquid IS Industrial Symbiosis LP Linear Programming LUR Lower Uncertain Region MEN Mass Exchange Network MILP Mixed-Integer Linear Programming MINLP Mixed-Integer Nonlinear Programing MSA Mass Searating Agent NLP Nonlinear Programming NPV Net Present Value NREL National Renewable Energy Laboratory OA Outer Aroximation ODP Ozone Deletion Potential xxi

24 List of Nomenclature ORC Organic Rankine Cycle PCP Perturbed Chain Polar PSE Process System Engineering POPC Palm Oil Processing Comlex QSAR Quantitative Structure-Activity Relationshis QSPR Quantitative Structure-Proerty Relationshis SA Simulated Annealing SAFT Statistical Associating Fluid Theory SCP Sustainable Consumtion and Production THFA Tetrahydrofurfuryl alcohol TI Toological Index TS Tabu Search UNIFAC Universal Quasichemical Functional-Grou Activity Coefficients UR Uncertain Region UUR Uer Uncertain Region VR Variable Range Indexes b index for biomass feedstock xxii

25 List of Nomenclature d index for signature i index for first order molecular grou for grou contribution method j index for second order molecular grou for grou contribution method k index for third order molecular grou for grou contribution method m index for chemical comonent in mixture index for target roduct roerty q index for conversion athway for biomass in integrated biorefinery q index for conversion athway for intermediate in integrated biorefinery s index for intermediate in integrated biorefinery s index for roduct in integrated biorefinery y index for individual objective function for weighted sum multiobjective otimisation aroach Parameters/Variables weighted sum A overall objective function for weighted sum multi-objective otimisation aroach A individual objective function for weighted sum multi-objective otimisation aroach xxiii

26 List of Nomenclature a weighting factor for A for weighted sum multi-objective otimisation aroach Bio B b flow rate of biomass feedstock b CRF caital recovery factor C i contribution of first order molecular grou C j contribution of second order molecular grou C k contribution of third order molecular grou Bio E b cost of biomass feedstock b Prod E s' cost of roduct s Ca E bq caital cost for conversion of biomass feedstock b through athway q Ca E sq' caital cost for conversion of intermediate s through athway q' Or E bq oerating cost for conversion of biomass feedstock b through athway q Or E sq' oerating cost for conversion of intermediate s through athway q I F bq flow rate of biomass feedstock b to athway q II F sq' flow rate of intermediate s to athway q G molecular grah GP Total total annualised cost xxiv

27 List of Nomenclature h height of signature H v heat of vaorisation HHV higher heating value K oc soil sortion coefficient l colouring sequence for molecular signature descritor L TI values for each signature root LC 50 lethal concentration/toxicity LFL lower flammability limit MW molecular weight n 1 number of signatures of valency one n 2 number of signatures of valency two n 3 number of signatures of valency three n 4 number of signatures of valency four N i number of occurrence of first order molecular grou N j number of occurrence of second order molecular grou N k number of occurrence of third order molecular grou N S total number of signatures in a molecule N DB number of signatures with one double bond in a molecule N MB number of signatures with two double bonds in a molecule xxv

28 List of Nomenclature N TB number of signatures with one trile bond in a molecule O number of circuits in a molecular grah OC oxygen content R retention R 2 coefficient of determination RON research octane number I R bqs conversion rate of biomass feedstock b to intermediate s II R sq's' conversion rate of intermediate s to roduct s TAC total annualised cost TACC total annualised caital cost TAOC total annualised oerating cost TLC toxic limit concentration T b boiling oint T ig auto-ignition temerature T f flash oint Inter T s total roduction rate of intermediate s Prod T s' total roduction rate of roduct s u constant secific to TI xxvi

29 List of Nomenclature UFL uer flammability limit L v lower limit for target roerty MIX, L v lower limit for target roerty for mixture LL v lower lower limit for target roerty LU v lower uer limit for target roerty U v uer limit for target roerty MIX, U v uer limit for target roerty for mixture UL v uer lower limit for target roerty UU v uer uer limit for target roerty V E affinity V value for target roerty MIX V value for target roerty for mixture V * value for target roerty with unknown target roerty ranges VP vaour ressure V m molar volume w I, w II, w III, w IV, w V, w VI correlation constants for grou contribution models x m fraction of chemical comonent m in mixture xxvii

30 List of Nomenclature z I binary coefficient of second order molecular grou z II binary coefficient of third order molecular grou α occurrence number of each signature χ connectivity index χ v valence connectivity index Δχ delta connectivity index χ n connectivity index of molecular non-olar structure δ Hildebrand solubility arameter ε edge adjacency index degree of satisfaction for target roerty r degree of satisfaction for roerty robustness for target roerty s degree of satisfaction for roerty sueriority for target roerty * degree of satisfaction for target roerty in two-hase aroach μ mobility η viscosity Ω normalised roerty oerator for target roerty xxviii

31 List of Nomenclature Ω m normalised roerty oerator for target roerty for chemical comonent m MIX Ω normalised roerty oerator for target roerty for mixture L Ω lower limit for normalised roerty oerator Ω MIX, L Ω lower limit for normalised roerty oerator MIX Ω U Ω uer limit for normalised roerty oerator Ω MIX, U Ω uer limit for normalised roerty oerator MIX Ω ζ standard deviation for roerty rediction model for target roerty θ roerty function corresonding to target roerty xxix

32 Chater 1 CHAPTER 1 INTRODUCTION 1.1. Problem Background Since the age of industrialisation, fossil fuels (etroleum, natural gas and coal) have been widely utilised as the main source of energy for heat and ower roduction. Besides heat and ower, fossil fuels have also been an imortant feedstock for the roduction of various commodities and secialty chemicals. Although the utilisation of fossil fuels has rovided develoments and conveniences to the society, the extensive consumtion of fossil fuels has also contributed to a number of environmental issues (Panwar et al., 2011). One significant environmental issue is the increased level of greenhouse gases (GHGs) in the atmoshere. In addition, the vast consumtion of fossil fuels has also led to the diminishing of fossil fuel reserves (Farhad et al., 2008). In order to simultaneously mitigate climate change and reduce the deendence on fossil fuels in transort and industry sectors, alternative source of energy is necessary. Such imlications have driven the industrial and scientific community to search for a more sustainable and renewable source of energy (Cherubini, 2010). As such, biomass has been identified as a otential alternative to fossil fuel resources as a feedstock for industrial roductions, addressing both energy and non-energy sectors (Naik et al., 2010). 1

33 Chater 1 According to Demirbas (2009), biomass comrises biological materials which can be found in natural and derived materials. It can be defined as organic matters in which solar energy is stored. Biomass contains varying amounts of cellulose, hemicellulose and lignin which are made u from hydrogen, carbon, oxygen, nitrogen and sulhur which resents in minor roortions. Saxena et al. (2009) categorises biomass resources into three main categories, which exist deending on different geograhic conditions. Energy cros which include edible and non-edible cros Standing forest which comrises various intermediate roducts and residual wastes of different nature Wastes which consists of agricultural, agro industrial and municial solid wastes Since ancient times, biomass has always been a major source of energy for mankind. Traditionally, biomass had been utilised through direct combustion. At resent, biomass can be converted into three main roducts of heat/electrical energy, transortation fuels and feedstock for chemical roduction. In recent decades, biomass utilisation as a source of energy is exeriencing a huge usurge in commercial and research interest due to several reasons. First of all, biomass is a renewable, otentially sustainable and relatively environmentally benign source of energy (Yuan et al., 2013). In addition, biomass can be converted into a huge array of diverse materials and value-added roducts. It rovides rooms for exloitation to be exlored and converted into different forms and roducts (Bozell, 2008). Moreover, the 2

34 Chater 1 combustion of forestry and agricultural biomass for energy roduction is an effective use of waste roducts. Hence, waste disosal roblem can be significantly reduced (Saxena et al., 2009) Integrated Biorefinery Biomass can be converted into useful forms of energy and value-added roducts by using a number of different rocesses. Presently, various wellestablished standalone biomass conversion rocesses and technologies are available to convert biomass into different value-added roducts. These conversion rocesses include hysical/mechanical, thermochemical, chemical and biological/biochemical rocesses. Biorefinery, a concet analogous to etroleum refinery can be alied to exlain the utilisation of biomass as a feedstock to roduce multile roducts (Fernando et al., 2006). According to Kamm et al. (1998), biorefinery can be defined as a comlex system made u from sustainable and environmentally benign technologies to exloit and utilise biological raw materials. Similar to a etroleum refinery, a biorefinery integrates a range of rocessing technologies to roduce multile roducts by using various biomass resources as raw materials. The imlementation of the biorefinery concet has been a key to chemical industries gradual shift towards the use of biomass resources for industrial chemical manufacturing (Frost and Draths, 1995). As biomass is available in different forms and characteristics, many ossible rocessing technologies are available to convert biomass into different value-added roducts. Hence, significant effort in the research and develoment for the synthesis and design of biorefineries has been focused on integrating the broad range of biomass conversion latforms 3

35 Chater 1 (hysical/mechanical, thermochemical, chemical and biological/biochemical conversion rocesses) to roduce a wide sectrums of value-added roducts. In order to enhance the roductivity, cost effectiveness and overall rocess erformances, Fernando et al. (2006) roosed the concet of integrated biorefinery which integrates multile biomass conversion latforms as a whole. Through the concet of integrated biorefinery, the waste generated from a biorefinery can be minimised while the energy and material recovery can be maximised. Hence, the concet of integrated biorefinery has the otential to imrove economic, environmental and social sustainability while converting biomass into energy and different chemical roducts efficiently. To date, there are a large number of established biomass conversion athways available for imlementation in an integrated biorefinery to convert biomass into a wide sectrum of otential roducts. As such, there is a need to screen all otential athways systematically and determine the otimum conversion athways that roduce the desired roducts. Various aroaches have been develoed for rocess synthesis and screening of otential conversion athways for integrated biorefineries (Bao et al., 2011; Halasz et al., 2005; Ng et al., 2009; Ng et al., 2014; Santibañez-Aguilar et al., 2011; Wang et al., 2013; Zhang et al., 2014; Zondervan et al., 2011). While significant amount of research attention have been focused in the area of synthesis and design of integrated biorefinery, limited research works have been carried out in the area of roduct design in integrated biorefinery. According to Skibar et al. (2009), chemicals derived from biomass have the otential to form the foundation of a huge range of roducts coming from 4

36 Chater 1 chemical industries. The number of otential value-added chemicals, either in the form of end roducts or as intermediate roducts is tremendous (Wery and Petersen, 2004). In addition to being utilised to roduce the conventional and market available roducts (i.e. bioolymers, biodiesel, solvents etc.), biomass has the ability to be designed and converted into various novel chemical roducts (Holladay et al., 2007). Thus, in order to convert biomass into value-added roducts efficiently, roduct design asect is essential to be taken into consideration in the synthesis and design of integrated biorefinery. This can only be achieved by integrating the design of integrated biorefinery with chemical roduct design Chemical Product Design A chemical roduct can be defined as a system consists of different chemical substances which is designed and manufactured for one or more uroses (Cisternas and Gálvez, 2006). According to Cussler et al. (2010), chemical roducts can be generally categorised into three tyes. The first tye is commodities such as acids and alcohols. Most of these chemical roducts are relatively simle to roduce, and the manufacturing rocesses involved are stable and well-established. Therefore, the design goal of commodities is to manufacture roducts at minimum cost. The second tye of chemical roducts is molecular roducts such as harmaceutical drugs. The selling oint for this tye of roducts is the rate of innovation and introduction of the roducts into the market. In order to comete with the rival comanies, the seed in discovering and develoing the roducts is more vital than the manufacturing cost of the roducts. The third tye of chemical roducts is erformance 5

37 Chater 1 roducts. The value of this tye of roducts is deendent on its functions, which are normally defined by the structure of the roducts. Table 1.1 summarised the classification of chemical roducts. Table 1.1: Classification of chemical roducts Commodities Molecular roducts Performance roducts Design key Cost Seed Function Design basis Unit Oerations Chemistry Microstructure Design risk Feedstock Discovery Science Although these distinct categories of chemical roducts differ in their design key stes, selling oints and ossible risks encountered during the roduct design rocess, the design rocedures for these chemical roducts are identical and similar. According to Cussler and Moggridge (2001), the main urose of chemical roduct design is to identify the otimal roduct to be made for a secific alication. Moggridge and Cussler (2000) roosed that the entire chemical roduct design rocess can be reresented by four rincial stes as follows: Identify customer needs Generate ideas to meet the needs Select among ideas Manufacture roducts 6

38 Chater 1 The first ste is to define the needs or functionalities that the roduct should fill. This is followed by generating all ossible ideas to meet the identified needs and selecting the most romising one among the generated ideas. The final ste in the design rocess is to determine the ways and methods to manufacture and test the roduct before introducing it to the market. Traditionally, bottom-u aroaches are used in searching for new chemical roducts with otimal erformance. For examle, a solvent is designed to rovide maximum searability during searation rocess. Bottom-u roduct design aroaches are usually based on design heuristics, exerimental studies and exert judgements or exeriences (Odele and Macchietto, 1993). In bottom-u aroaches, all feasible molecules are first generated from the raw materials and subsequently tested for the required roduct needs. As they are mainly based on trial and error aroaches, these traditional methods are intrinsically inefficient, time consuming and costly (Venkatasubramanian et al., 1994). Furthermore, these aroaches are largely deendent on the available information and knowledge. Thus, it is challenging to search for new chemical roducts which ossess otimal roerties without systematic selection tools (Churi and Achenie, 1996). On the other hand, new chemical roducts can be designed via to-down aroaches. To-down aroaches are reverse engineering aroaches where the design rocess begins with the identification of needs to fulfil, and search for the molecules that ossess the roerties which can meet the roduct needs (Gani et al., 1991). 7

39 Chater 1 As discussed earlier, the goal of chemical roduct design is to identify a roduct that gives a articular erformance. In most cases, erformance and suitability of a roduct are defined in terms of hysical roerties rather than chemical structure of the roduct. For examle, to design an effective refrigerant, the volumetric heat caacity for the designed refrigerant should be high so that the amount of refrigerant required is reduced for the same refrigeration duty. Besides, the viscosity of the designed refrigerant should be low to achieve a low uming ower requirement. Hence, as long as the roduct ossesses high volumetric heat caacity, low viscosity and fulfils other roduct needs, it is suitable to be used as an effective refrigerant regardless of its chemical structure. Therefore, chemical roduct design roblems can be considered as inverse roerty rediction roblems where the referred roduct attributes are reresented in terms of hysical target roerties. The objective for the inverse roerty rediction roblems is to determine the molecule that matches the defined roerties (Gani and O Connell, 2001). As stated by Stehanooulos (2003), one of the imortant sources of roduct secifications and requirements in roduct design is customer needs. Thus, it is required to translate descritive customer requirements into measurable hysical roerties of a roduct (Achenie et al., 2003). For examle, in order to design a chemical roduct which is non-hazardous, toxic limit concentration of the roduct should be measured. Another examle would be the design of a roduct which will not cool easily. In order to fulfil the requirement, the heat caacity of the roduct is measured and taken into 8

40 Chater 1 consideration during the design rocess. In addition, there are situations where a roduct needs can only be fulfilled by measuring and taking more than one hysical roerty into consideration; for examle, design of a transortation fuel. The consistency of the fluid flow has to be considered to make sure that the fuel can flow continuously from the fuel tank to the combustion chamber. Hence, the density and viscosity of the fuel have to be considered during the design stage. Moreover, engine efficiency is one of the imortant arameters which should be considered in designing the fuel so that the fuel can be burnt to run the engine efficiently. Based on such requirement, octane rating and heating value of the fuel are taken into account. After the required roduct attributes are reresented with measurable roduct roerties, chemical roduct that meets the roduct needs can be designed based on the identified roduct roerties. The design of chemical roducts based on roduct roerties can be done by using comuter-aided molecular design (CAMD) techniques. In general, CAMD techniques redict and estimate the roerties of molecules by using roerty rediction models. According to Gani and Pistikooulos (2002), roerty rediction models estimate roduct roerties from structural descritors, which are numerical values that contains chemical information of a molecule. Some of the commonly used structural descritors to quantify a molecular structure include chemical bonds and molecular geometry (Randić et al., 1994). As CAMD techniques are able to design chemical roducts having a set of desirable roduct roerties, they have been develoed as owerful techniques in the field of chemical roduct design 9

41 Chater 1 (Harer and Gani, 2000). Currently, most of the CAMD techniques utilise roerty rediction models based on grou contribution (GC) methods to verify that the generated molecules ossess the secified set of target roerties (Harer et al., 1999). Besides GC methods, established methods in develoing roerty rediction models include the alication of toological indices (TIs). TIs are used to correlate the chemical structure to hysical roerties of a molecule. The correlated relationshis are called quantitative structure roerty/activity relationshis (QSPR/QSAR) (Kier & Hall, 1986). To date, most of the develoed and available chemical roduct/molecular design aroaches emhasise on designing otimal chemical roducts/molecules. In most cases, the otimal roduct is designed in terms of otimum target roerty(s). This otimality of roduct roerties is the main factor that defines the quality of a roduct. Hence, in inverse roerty rediction chemical roduct design roblems, the roduct with otimal redicted target roerty(s) will be regarded as the otimal roduct. As mentioned earlier, most of the CAMD techniques involve roerty redictions which are done by using roerty rediction models. It is noted that roerty rediction models are develoed with certain accuracy and uncertainties. The accuracy of these roerty rediction models are usually exressed in terms of correlation coefficient (r 2 ) and standard deviation (ζ). In general, the accuracy of roerty rediction models is used as an indicator of the models ability in redicting the roerties or the exected error that the models might cause. Hence, while utilising CAMD techniques in designing 10

42 Chater 1 chemical roducts with otimal redicted target roerty(s), the exected accuracy can be known. In addition, most of the established chemical roduct design methodologies are focusing on otimising a single roduct roerty while designing an otimal chemical roduct. It is aware that in some situations, there are several imortant roduct roerties to be considered and otimised in order to design an otimal chemical roduct. For examle, during the design of an effective solvent, the solvent should be designed with maximum searability or solubility. Besides, the flammability limit of the designed solvent should fall within the safety oerating limit and the toxicity of the solvent should meet the environmental regulations (Harini et al., 2013). Refrigerant design is another examle where multile roduct roerties are imortant. In order to design an effective refrigerant, the volumetric heat caacity for the designed refrigerant should be high so that the amount of refrigerant required is reduced for the same refrigeration duty (Samudra and Sahinidis, 2013). Besides, the designed refrigerant should ossess low viscosity to achieve low uming ower requirement. Since more than one design objective is involved in designing these roducts, these chemical roduct design roblems have to be solved as multi-objective otimisation roblems. In many occasions, chemical roducts exist in the form of mixtures of different comonents rather than as single comonent roducts. According to 11

43 Chater 1 Churi and Achenie (1997), the main urose of designing mixtures is that mixtures have the otential for giving a good mix of target roerties which is unattainable by individual chemical comonents. For examle, for refrigerant design, although difluoromethane (CH 2 F 2 ) is a better refrigerant in terms of low comressor dislacement and zero ozone deletion, its vaour ressure is too high for an efficient refrigerant. Therefore, in order to overcome this limitation, mixture of chloromethane (CH 3 Cl) and CH 2 F 2 is introduced as mixtures offer greater flexibility in matching multile target roerties (Duvedi and Achenie, 1997). As ointed out earlier, various biomass conversion athways are available to convert biomass into a wide range of roducts. In cases where the desired roerties cannot be met by a single comonent roduct, an otimal mixture of chemicals would be required Problem Statement It is realised that the synthesis and design of integrated biorefineries which focus on the roduct design asects are yet to be examined thoroughly. Many established aroaches have been develoed by the research community in the realm of integrated biorefinery. However, most of the available aroaches emhasise on the identification of otimal rocessing routes that lead to the roducts without addressing the roduct design asects of the biorefinery. This serves as the main motivation in this thesis. In chemical roduct design roblems, there are cases where multile roduct roerties are required to be considered and otimised to design an otimal roduct. In such cases, the chemical roduct design roblems have to be solved as multiobjective otimisation roblems. Moreover, roerty rediction models are 12

44 Chater 1 develoed with certain range of accuracy. The accuracy of roerty rediction models is necessary to be taken into consideration while utilising CAMD techniques for chemical roduct design roblems. On the other hand, while converting biomass into value-added roducts in an integrated biorefinery, there are situations where the desired roduct needs cannot be fulfilled by a single comonent roduct. In such situations, design of otimal mixture from biomass would be required. All of the abovementioned remain as research gas to be filled within this thesis Research Objectives Based on the identified research gas, the rimary objectives of this research work can be summarised as follows. i. To develo a concetual framework for the identification of chemical roduct and its roduction routes in integrated biorefineries. ii. To develo a systematic methodology for the design of otimal chemical roduct in terms of target roduct roerties as well as the identification of otimal conversion athways based on different otimisation objectives in integrated biorefineries. iii. To develo a multi-objective otimisation aroach for otimal chemical roduct design roblems where multile roduct roerties are required to be considered and otimised simultaneously. 13

45 Chater 1 iv. To develo a robust chemical roduct design methodology that considers the accuracy of roerty rediction models while designing an otimal chemical roduct in terms of target roduct roerties. v. To develo a systematic methodology for otimal mixture design in terms of multile roduct roerties via otimal conversion athways based on different otimisation objectives Outline of Thesis In this thesis, introductions to integrated biorefinery and chemical roduct design are resented in Chater 1. In addition, Chater 1 also highlights the main research ga and the objectives of this research work. Chater 2 resents a critical and thorough literature review on the otential chemical roducts that can be roduced from biomass, different CAMD techniques as well as the current status in the area of CAMD. This is followed by the discussion of research scoes and research methodologies, which are resented in Chater 3. Chaters 4 8 discuss various novel integrated design techniques for biorefineries with emhasis on chemical roduct design asects. A concetual design for the synthesis of chemical roduct from biomass in integrated biorefineries is resented in Chater 4. The objective for the concetual design is to identify chemical roduct from biomass and its roduction routes in integrated biorefineries. Chater 5 extends the concetual design into a 14

46 Chater 1 systematic methodology for otimal chemical roduct design in integrated biorefineries. The systematic methodology designs the otimal chemical roduct in terms of target roduct roerties. After the design of the chemical roduct, the otimal conversion athways based on different roduction objectives in converting the biomass into the chemical roduct are identified. Chater 6 resents a multi-objective otimisation aroach for the design of otimal chemical roducts in terms of multile roduct roerties. The resented aroach rovides solutions for chemical roduct design roblems where multile roduct needs are required to be considered and otimised simultaneously. In chater 7, a robust chemical roduct design aroach is introduced. The aroach identifies otimal chemical roduct in terms of otimality of the roduct roerties while considering roerty rediction model accuracy. Chater 8 further extends the systematic methodology resented in Chater 5 for the design of otimal mixture from biomass which meets the roduct needs as well as the identification of otimal conversion athways in roducing the mixture in terms of different roduction objectives. Chater 9 resents the summary of the accomlishments and contributions of this research work. In addition, otential future works that can be done based on the techniques develoed in this research work are highlighted in this chater. 15