PONTIFICIAUNIVERSIDADCATOLICADECHILE SCHOOLOFENGINEERING MANAGING UNCERTAINTY IN AGROFORESTRY PROBLEMS: APPLICATIONS OF OPERATION RESEARCH MODELS AND METHODOLOGIES IN THE WINE AND FORESTRY INDUSTRIES ´ ´ MAURICIO ANDRES VARAS VALDES ThesissubmittedtotheOfficeofGraduateStudies inpartialfulfillmentoftherequirementsforthedegreeof DoctorinEngineeringSciences Advisor: SERGIOMATURANAVALDERRAMA SantiagodeChile,January2016. (cid:13)c MMXVI,MAURICIOANDRE´SVARASVALDE´S. PONTIFICIAUNIVERSIDADCATOLICADECHILE SCHOOLOFENGINEERING MANAGING UNCERTAINTY IN AGROFORESTRY PROBLEMS: APPLICATIONS OF OPERATION RESEARCH MODELS AND METHODOLOGIES IN THE WINE AND FORESTRY INDUSTRIES ´ ´ MAURICIO ANDRES VARAS VALDES MembersoftheCommittee: SERGIOMATURANAVALDERRAMA ALEJANDROMACCAWLEYVERGARA JORGEVERAANDREO V´ICTORALBORNOZS. SUSANCHOLETTE CRISTIANVIALEDWARDS ThesissubmittedtotheOfficeofGraduateStudies inpartialfulfillmentoftherequirementsforthedegreeof DoctorinEngineeringSciences SantiagodeChile,January2016. Toallthepeoplethat,inonewayor another,madethisthesispossible. ACKNOWLEDGEMENTS To Professor Sergio Maturana, my Thesis Advisor, for giving me the honor of work- ing with him. For all his kindness, his great enthusiasm in the development of this thesis, for his unconditional academic support in all the fields of management sciences; but most importantly,forteachingmehowtobearighteousman. Withoutbothhisacademicencour- agementandrigorousness,thisresearchwouldhavebeenimpossibletobeaccomplished. To Professor Jorge Vera, for helping me with all my mathematical inquiries and for showing me the beauty of operations research. Also, I want to thank him for showing me howthepassionforworktranslatesintoteachingexcellence. To Professor Alejandro Mac Cawley, for giving me the chance to work with him as a teacher assistant and learn from him the appeal of operations management. Besides, I also want to thank him for his support and for encouraging me to finish my dissertation when I thoughtitwasanimpossibletask. To Jacqueline Lefin, secretary of the Industrial and Systems Engineering Department, and Pilar Mart´ınez, administrative coordinator of the same Department, for all their kind- ness and support since the undergraduate program. Also, I want to acknowledge Angela Saba´ and Marcela Berr´ıos, secretaries of the Industrial Engineering Master Program at the MII. TomyfriendsfromboththeUndergraduateandGraduateProgram,whomademystay at the engineering school and the doctoral program pleasant. Special thanks to Claudio Araya, Pamela Carrillo, Alejandro Cataldo, Carolina Chamorro, Raimundo Elicer, Armin Lu¨er, “Dino” Ortega, Walter Quintana, Raimundo Steidle, Ignacio Vargas and Cristian Villalobos. Thanks for your friendship, your continuous support, and for sharing with me yourincrediblewisdom. And finally, to my family, for giving me hope in those moments when I thought I was not capable of carrying out this research. Particularly, Ce´sar, my father, has been my iv exampleofbothresilienceandself-improvementallmylife. AllthatIamisbecauseofyou. To my Mother Nancy, my brother Pablo, and my beloved sister Karina (R.I.P.) for giving me their support and for always being there for me. To Katherine, my wife, who faced the hard task of taking care of our son at the times when I was absent for the requirements of the doctoral program. I will be forever grateful, Kath. And to Clemente, my firstborn; everythingIdo,Idoitthinkingaboutyou. Last, but not least, this thesis was supported by FONDECYT grants 1110367 and 1150882,andspeciallybyMECESUPgrantPUC0710. v TABLEOFCONTENTS ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv LISTOFTABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix LISTOFFIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi 1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1. Operationresearchmodelsinforestry . . . . . . . . . . . . . . . . . . . 1 1.1.1. Theforestproductionchainandtheroleofsawmills . . . . . . . . . . 3 1.1.2. Thesawmillproductionschedulingproblem . . . . . . . . . . . . . . 5 1.1.3. The impact of uncertainty on the development of feasible production schedules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.1.4. Researchonthesawmillproductionproblemunderuncertainty . . . . 6 1.1.5. The scope of scheduling production for a sawmill: a robust optimization approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.2. Operationresearchmodelsinagriculture . . . . . . . . . . . . . . . . . . 8 1.2.1. Theagri-foodsupplychainandthecaseofwine . . . . . . . . . . . . 11 1.2.2. ORmodelsappliedinthewineindustry . . . . . . . . . . . . . . . . 15 1.2.3. Theimpactofuncertaintiesinwineproductionactivities . . . . . . . 17 1.2.4. Themanagementofhigh-valuegoodsinthewineindustryunderuncertainty 19 1.2.5. Lot-sizing in an export-focused winery and the impact of postponement practicesunderuncertainty . . . . . . . . . . . . . . . . . . . . 20 1.3. Objectiveandmainhypothesisofthisthesis . . . . . . . . . . . . . . . . 21 1.4. Theremainderofthisthesis . . . . . . . . . . . . . . . . . . . . . . . . . 21 2. SCHEDULINGPRODUCTIONFORASAWMILL:AROBUSTOPTIMIZATION APPROACH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.2. Robustoptimizationmethodology . . . . . . . . . . . . . . . . . . . . . 27 vi 2.3. Theschedulingproductionmodel . . . . . . . . . . . . . . . . . . . . . . 29 2.3.1. Problemdescription . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.3.2. Modelformulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.4. Robustreformulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.4.1. Uncertaintyinproductdemand . . . . . . . . . . . . . . . . . . . . . 32 2.4.2. Uncertaintyinsupplyofrawmaterials . . . . . . . . . . . . . . . . . 34 2.5. Computationalanalysisofthemodel . . . . . . . . . . . . . . . . . . . . 37 2.5.1. Uncertaintyindemandparameters . . . . . . . . . . . . . . . . . . . 37 2.5.2. Uncertaintyinsupplyparameters . . . . . . . . . . . . . . . . . . . 44 2.5.3. Uncertaintyinbothdemandandsupplyparameters . . . . . . . . . . 47 2.6. Concludingremarksandfutureresearch . . . . . . . . . . . . . . . . . . 54 Appendix2.A:thedata . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 3. MANAGING PREMIUM WINES UNDER UNCERTAINTY: A TWO-STAGE HEURISTICTOSETTINGSTOCKLEVELS . . . . . . . . . . . . . . . . . . 58 3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.1.1. Managing high-value goods in the wine industry under uncertainty: preliminaryconsiderations . . . . . . . . . . . . . . . . . . . . . 58 3.2. Computing the mean delays of customer orders for an asymmetric cyclic queuesystemunderanexhaustiveservicepolicy . . . . . . . . . . . . . . . . . 64 3.2.1. Winandsetal. approach . . . . . . . . . . . . . . . . . . . . . . . . 64 3.2.2. Anumericalexample . . . . . . . . . . . . . . . . . . . . . . . . . . 69 3.3. Settingheuristicallywinestocklevelsusingan(s-1,s)inventorycontrolpolicy 73 3.3.1. Theoptimizationproblem . . . . . . . . . . . . . . . . . . . . . . . 73 3.3.2. Solvingthewinestocksettingproblem . . . . . . . . . . . . . . . . 75 3.3.3. Anumericalexample(cont.) . . . . . . . . . . . . . . . . . . . . . . 81 3.4. Theheuristicaccuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 3.4.1. Theheuristicaccuracyinourworkingexample . . . . . . . . . . . . 83 3.4.2. Thesymmetricalcase . . . . . . . . . . . . . . . . . . . . . . . . . 84 3.5. ConcludingRemarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 vii Appendix3.A:Proposition3.2 . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Appendix3.B:Lemma3.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 4. A ROLLING HORIZON APPROACH TO ANALYZE THE EFFECTS OF POSTPONEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 4.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 4.2. Overviewofwineryproductionoperations . . . . . . . . . . . . . . . . . 91 4.3. Modelsforwineproductionplanning . . . . . . . . . . . . . . . . . . . . 94 4.3.1. Modelwithpostponementinthelabelingprocess . . . . . . . . . . . 95 4.3.2. Abenchmarkmodelwithoutpostponementinthelabelingprocess . . 98 4.4. Experimentalanalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 4.4.1. Aframeworkfordecisionmakingandtheexperimentalscope . . . . . 101 4.4.2. Sets and instance parameters, rolling horizon setup and performance measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 4.4.3. Capacitytightnessandcertaindemanddatageneration . . . . . . . . 105 4.4.4. Error-pronedemandforecasting . . . . . . . . . . . . . . . . . . . . 106 4.4.5. Experimentalresults . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4.5. Concludingremarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 5. CONCLUSIONSANDFUTURERESEARCH . . . . . . . . . . . . . . . . . 114 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 APPENDIXA. ANOTEONTHENEWSVENDORPROBLEMUNDERPOISSON DEMAND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 A.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 A.2. Settingtheoptimalorderquantityandbounds . . . . . . . . . . . . . . . 132 A.3. Behavioroftheoptimalsolutions∗(λ) . . . . . . . . . . . . . . . . . . 135 A.4. Behavioroftheoptimalvalueh∗(λ) . . . . . . . . . . . . . . . . . . . . 138 A.5. ConcludingRemarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 viii LISTOFTABLES 1.1 Taxonomyforcitedpapers. . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1 SetsusedintheSchedulingProductionModel . . . . . . . . . . . . . . . . . 30 2.2 VariablesusedintheSchedulingProductionModel . . . . . . . . . . . . . . 30 2.3 ParametersusedintheSchedulingProductionModel . . . . . . . . . . . . . 31 2.4 GeneralLot-SizingandSchedulingProblemNotation . . . . . . . . . . . . . 31 2.5 Average operationalcosts forselected values ofraw materialsupply and product demandconservatismparameters. . . . . . . . . . . . . . . . . . . . . . . . 54 3.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 3.2 Optimalsolutionsandvalues. . . . . . . . . . . . . . . . . . . . . . . . . . 83 3.3 Stationary distribution estimation of the number of order waiting or being processed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 3.4 Heuristicinventorypositionforeachqueueinthesymmetriccase. . . . . . . 85 3.5 Estimationofthestationarydistributionatthesymmetricalcaseafterareplication lengthof10,000,000hours. . . . . . . . . . . . . . . . . . . . . . . . . . . 85 3.6 Comparisonbetweentheheuristicandtheoptimalsolution. . . . . . . . . . . 85 4.1 Setsandparameterinstancevalues . . . . . . . . . . . . . . . . . . . . . . 103 4.2 Computation of the performance measures (average per period) for the model whichallowspostponement. . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.3 Computation of the performance measures (average per period) for the model withoutpostponement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.4 Averageperperiodperformanceofimplementeddecisionsforγ =0.1 . . . . 108 4.5 Statisticalanalysisforthedifferenceswithγ =0.1 . . . . . . . . . . . . . . 108 4.6 Averageperperiodperformanceofimplementeddecisionsforγ =0.5 . . . . 109 ix 4.7 Statisticalanalysisforthedifferencesatγ =0.5 . . . . . . . . . . . . . . . . 109 4.8 Averageperperiodperformanceofimplementeddecisionsforγ =1.0 . . . . 109 4.9 Statisticalanalysisforthedifferencesatγ =1.0 . . . . . . . . . . . . . . . . 109 A.1 Rootsofthetranscendentalequations(A.19)forl ∈ {0..10}. . . . . . . . . . 137 (cid:16) (cid:17) A.2 Slopeofthelinearinterpolants m = h∗(λl)−h∗(λl−1) joiningconsecutivepoints l λ−λ l l−1 ofsetP = {(λ ,h∗(λ )) : l ∈ Z+ ∪{−1}},for(cid:15) = {3, 1}. . . . . . . . . . . 139 l l 4 4 x
Description: