Loading...
HomeMy WebLinkAboutRisk-Based Forecasts of Natural Gas Supplies for the Alaska Railbelt Regional Integrated Resource Plan Decision Tree Model 1.0 Sept 22, 2009Natural Gas Supplies for the Alaska Railbelt Regional Integrated Resource Plan (RIRP) Decision Tree Model 1.0 a a Black &Veatch Corp. Enterprise Management Solutions September 22,2009 PERROTTA ET:as 3 EE REE TI.Ld BPCUILDINGN WORLD,:aoRENCESlgeSoSBLACK&VEATCHFuel-'Supply Task for RIRP ©Review outlooks for the major hydrocarbon fuels that historically have been utilized in the Alaska Railbelt region for electric power generation and other purposes Natural gas HAGO (Heavy Atmospheric Gas Oil;Fuel Oil #4) Naphtha Coal Propane (addition to scope proposed by a stakeholder) eo For each fuel,as delivered at Cook Inlet,provide annualized forecasts for a 50-year timeline of 2011-2060 Supply (MMBtu or other volume per year) Price ($US /MMBtu) ERSTE RENEEEITATa aE Saane aeRO ER,a HSN Page -2 pe Re ghee ORY Wag ATO ESCA 'Ss OE SERRE rR SE Sars "4"Draft Work Product.="Not.for Publication Sep 22,2009 | Fuel-Supply Task for RIRP Fuel-Supply Task for Regional IRPForecastOutput D,=Deterministic (annualized average) Fuel P,,,=Probabilistic (xx percentage probability of attainment) Supply*Price* (MMBtu/d)($/MMBtu)Analytics e Decision-tree model Natural gas -Pigs Peo,Poo P19,P59;Poo |@ Matrix of scenarios and Monte Carlo simulations HAGO D,D,e Historical-empirical Naphtha D,D,models Coal D e Adjustments made for083Dsfundamentalsofeach Propane**D,D,fuel *Delivered at Cook Inlet AK **Proposed by stakeholder as an addition 7 "Page-3."Draft Work Product -_Not for Publication."Sep 22,2009 | 7 Caopaeenee Te TE gw le age -+ae :arian y we -.r te *.BNF -?7 4 ret”aa foe vey . we =s pee ee ys tm ON gilt 'eof, ans a Premise of Decision Tree Model for Natural Gas ©Decision Tree Combines Best of Two Forecast Worlds -Deterministic forecasts generally are simpler to make and easier to explain than are probabilistic forecasts” -But there is no single-path,deterministic forecast that can adequately capture all of the relevant uncertainties in natural gas supplies -Adecision tree combines the strength of a probabilistic forecast with the intuitive clarity of a deterministic forecast eo Decision Tree Provides Supply Outlook as Input to Price Forecast -Addresses supply-side uncertainties in the availability of natural gas -Allows different prospective supplies to be treated individually but also as part of a matrix of alternatives -Produces numerical likelihood values that a selected combination of sources will deliver certain benchmark gas volumes -Provides the "supply”input for a separate price forecast *Probabilistic forecasts often are called "risk-based”forecasts because they employ the elements of chance in calculating possible outcomes BRakeorg Fat PEMD aR FARMRS8d MESA PTD Es SORE ME ONO NE 5 r gc se Shs ARLESRSette RS.ge PR a(Page-4.s-s-s-sé'i@D EL.Work Product -Not forPublication...Sep22,2009} Natural Gas Decision Tree 1.0 Railbelt Energy Portfolio Portfolio Scenarios Base Case Growth Case :Demand y Supply-Source Supply-Management | Nodes .Nodes | Cook Inlet Gas Production Gas Storage Bullet Line System from 4 North Slope Spur Line Regional PipelinefromDistribution AGIA Trunkline System LNG Imports Gas from Coal AAOB opinttoneebing (eae Oe eee eee aa aea a eeeA Ce eee en eeSkec a aee et VvDraft.Work Product -Not forPublication_ Regional Gas Requirements AN SE RECLINESIE ART a,SER TBE eR ge a Sep 22,2009 ; JfaegEee<_F=Components of the -an |Gas Decision Tree 1.0Natura tiie Rupert Variable Type 1: ||]took miet Production|25.0%og Yes /No Availability [ Th 0 of Node | [Supply trom New Delivery f 4-08 Delive ty Sources-7 Bullet Line me <q Variable Typ e3:-+-4 -4 Pipeline from North Slope 0 %veuppy te v4 Q T=Spur Line from AGIA Trunkline I pase <j p-4 Railbett Gas Supply Pipeline Distribution System Coal from Gas ound ! 25.0%Variable Type 2:LNG Import ---e 0,Probability Distribution Function for Deliverable >| , Resources(-Outp ut:L |Geolagic Storage 100.0%<j -Total Gas Supply 50.0%/Stor, ©2020-2029 osm abonORE OT Supply from Storage :ne o [=Frequency -Cumulative %Pak a TT Ps =»+100%eae \fTank Storage |ar.aedaaer es 44 lh +50%BAP III I MRS eTIr oo] wep sss zu 0 Bett tty On,0% o 8 ee 50 200 400 600KLCapacity(MMcf /d)(_y *Built using Palisade PrecisionTree 5.0 Professional with @Risk Simulator,exported to Microsoft Excel 2003. RA Aper eetpage"6 "Draft Work Product -Not for Publication_iO Hee Rae BLACK&VEATCH A aa i ore ee TT"Sep22,2009 1?oo.ST"Components of the Natural Gas Decision Tree 1.0 aad eo Version 1.0:Benchmark for first 3 decades;vet with stakeholders -Choices must be made (see table below for example of one possible matrix) eo Version 2.0:Incorporate feedback and annualize for all 5 decades Risk-Based Natural Gas Forecast Strawman Simulation Matrix 1.0A @ =Primary node (major supply)©=Secondary node (minor supply) Annualized Gas Supply by Decade (P49,P59,Pog as MMcf/d)Model Node 2010-2019 j 2020-2029 |2030-2039 |2040-2049 |2050-2059 Cook Inlet Gas Production e e e@ O O Bullet Line from North Slope e @ O O O Spur Line from AGIA Trunkline O e e @ @ Gas from Coal O O O Oo O LNG Imports 7ee O O O O O Gas Storage Oe Cts O @ e e e Pipeline Distribution 'System @ @ @ @ @ 'Page-7-Ss- -s-s-s«éD aft Work Product -Not for Publication_"Sep 22,2009| feK&VEATCH. ut :-ay Tt |it wo .ene ..ae .a .pw ,acta RLD OF DIFFERENCE jceeeeeeeternenne meena an ee SLACPseeta”.-ast POS Ge weet oy 5 ere Bs _aan . Components of the Natural Gas Decision Tree 1.0 Construction of Probability Function for Each Node penneeeeeeeeeeee,'Probability ||Histogram by !|Functionby $! Decade ie,Decade ! Rank-Order |||2010-2019 |:'CX '|2010-2019 || Deliverability ||@Risk |Estimates by |_«t |_2020-2029 ]i |ci vare Fe L_2020-2029_|| Year ||2030-2039 ||[module |i |2030-2039 ||(MMcf/d)1 !!: ;||2040-2049 ||ee '|2040-2049 ||'|2050-2059 ||ne ||2050-2059 ||Gather Supply 1 ;I ; ee ee ee ee eeAvailabilityper|www!an |.Rene "Ly pe won Precision TreeposGyyucssoftware-Interviews OY a ue module Se i emer mie cena TeREE TaD I MLN Oe BRR MAREN UNEHEE RE Peataina ss IPReeFTGteegin nM|ene waren eer s CRP wot att 5 A CE ery 2 : .rag Pt stA ray wo EHP SPOR ARSENR IOSD /_Page-8 ||Draft Work Product -Not for Publication |_Sep 22,2009 | BLACK &VEATCHaRRB:s"Components of the Natural Gas Decision Tree 1.0 ©Cook Inlet Gas Production (Supply-Source Node) -Data Source:State of Alaska Department of Natural Resources (DNR,2009)-Assumptions: Physical production =deliverability of source-supply nodeSupply-Source Node:Cook Inlet Gas Production oe @ 2010-2019 -&-2020-2029 ++2030-2039 ProbabilityofAttainment-90 100 150 200 250 300 MMcf/d ORD MAE TIANFRE a RRORR REY A OF TYPES ST SAL OUT FOR POIRIER NOMEST ARaOR|Page-9 _"Draft Work "Product ---Not for Publication _"Sep 22,2009 |a OARFe GEN AE SRiya Pun esp 'BE ee TS tee ee git te ke .oa Se RE Alcase eres ae ke eet Pee raeREN®U CHEN FE Tee :ud f 4 ot eee a Components of the Natural Gas Decision Tree 1.0 eo Spur Line from AGIA Trunkline (Supply-Source Node) -Data Sources:AGIA reports (TransCanada base design) -Assumptions: Throughput builds at ramp rate quoted for AGIA Trunkline (in-service date distribution per Westney Consulting);first flow in 2017 Maximum deliverability is 500 MMcf/d when fully completed ProbabilityofAttainmentSupply-Source Node:Spur Line from AGIA Trunkline @ 2010-2019 -&--2020-2029 2 2030-2039 :::::**::'''Fit as Logistic Function woven Ghd vosedessabeeeeefeceecfeseecheeedeeeeefesceefessedheseedeeceefercecbassedeceefeseenfecconboeedieceefesceebegyobend webbed bdenbeberboeeddeenfeeboaebnfeeefenfecebenfen soe hon ee deneeefonecobeoeesbeoee|Siaaiatet ieteeie debate aemo neces boone pene edtewnae fovcacfecec spec encore edannee fees, serecedoc een foeceeponcecroree derecefeceenpeececones noose feooee!weredheeee foseee beceecbereecbenee qoreocfeee weve diccecbesecebessedieseedieneefeeeeebeceedieeeedeeeeefeceeEb sccb ccedbeedbececcheseefeecefecedipeanece AL Lleeefenefenendeffenhbo LJ t LI 200 300 400 500 MMef/d |Page-10.. s«dDraft'Work Product -Not for Publication.ES SAREE ERE ns aaAa Components c of the Natural Gas Decision Tree 1.0 e Bullet Line from North Slope (Supply-Source Node) -Data Sources:Inferences from ANGDA reports;Interview with State of Alaska -Assumptions: First flow in 2015;Builds at ramp rate derived from AGIA Spur Line forecast Maximum deliverability is 500 MMcf /d when fully completed Node is retired after full completion of Spur Line from AGIA Trunkline Supply-Source Node:Bullet Line -@ 2010-2019 -#-2020-2029 --2030-2039 (retirement assumption) 100%:'':':'H H H L H A i t t ::''' 90%toed fd dt AFit as Logistic Function}ii bb 80%4-0 70%4-1 60%4 50%- 40%- 30%4 20%+4 ls Gaal a EE a aE aoeoyfs -100 200 300 400 500 MMcf/d Le BRL HREOC DORpoe SR AIR RSM Osha"Page -1 a a "DraftWork.Product.-Not."for "Publication a Sep 22,2009 |ProbabilityofAttainmentAmenag!ET E08 aoe ee a Re cee eS prememeconanee meses -ee ;:7 2 s Ee,ok <BGSSaenoeweaBUILDING'A WORLD.OF,DIFFEREN CE case sennann nen aeant=wan nen Components of the Natural Gas Decision Tree 1.0 oe LNG Imports (Supply-Source Node) -Data Sources:B&V research;Interview with ENSTAR -Assumptions: Cargo receipt begins in 2013;1 cargo first year and maximum of 5 per year Each cargo is 138,000 cubic meters of LNG (=2.9 Bcf gas) Node is retired after full completion of Spur Line from AGIA Trunkline Supply-Source Node:LNG Imports -@ 2010-2019 -#-2020-2029 ++2030-2039 (retirement assumption) 100%''::' 90%ponncnneeeeneeeeeeeenfeceeeennee Fit as Logistic FUNCtION |J-----------en-feneene ee a a ron Pag gn ef 60%-|i en jovneen Le a psec a veeeeeeeceenenee joven40%feveneneneeeeneeee font eeetnseafssnMQ utatnnefnensennsnnne es en 30%4 |wr fenennnrnn frente Cees a A ae OL nner eeene 109%essesecenseeeeneere becentetntntntnfeetntnnne fgets oe ee 0%!3 i -20 40 MMef /d TEOMA ED ome,AE wa ce eee aa iE he eae 6 nk PS Sa1OOI Bay I POR MA RSA TE RTE One LORIN TEP RPP."Draft Work Product.-Not.for Publication _Sep 22,2009;ProbabilityofAttainmentsae Components of the Natural Gas Decision Tree 1.0 ©Gas from Coal (Supply-Source Node) -Data Sources:B&V research;Interviews with Usibelli Coal and PacRim Coal -Assumptions: -Coalbed methane produced at same rate as coal solids (i.e.,incidental rather than focused gas drilling and production)-so deliverability is negligible Supply-Source Node:Coalbed Methane -@®2010-2019 -#-2020-2029 100% 'Cr a rr rr a rs |og SNe EEE PE ENE PE pp EE bg90%-ieieieeienetaSS SE a RE DE a IRE REET Re SET SE OEE _WER RE GEE OEE OR SD RLforoog,a rr eC er i oto ooowvrrerLrSSSr,Ss cs | EF B0%oct ddd bebe beberle cbecdcdbecbenbeerbeerberdy H 4 :;:;8 'H H i oot '''14 ::nr)er 'E TO%onndenn bene deee eben dee bp bee beep ebbbd eb bd eeedeeedco.PoP db ob ob EON bb ob Pe bb ob rot Me bod bob cb ob gs oO :rr rr rn rr rr roe aS :::'sootOfahncdecenbenedecnchoccdecechonde.am.--fone cten wn been clennchacntennchane decnahenedaccchooPleccchecedeae-es eeetLeaatennateScieDaSsSAOSOEwui, r,S0,wh ce eee bee de cee cede cee nn dann!..oe...a oeNewenhawadewmabenetancabanadennnbenndpheubenedeneea er.ae rrdfGB50%poderbeterbecbebe rete edecbeeber de bebecnbecebeecbeeeeeefdiebeobootbonnbeendendood>Popo obo gd of bob ob iae gE oto dt goto topot tor,b bo obobobo i: a RE.SR.eeeSR"Gat ede De OR UacSPiiiiibEELELEEMEEbiii Pb pM i bbQO.20%qprbobebebepebepebebeAge beepeebeibenedeni10%+|Fit as Logistic Function].t.iMedi tb tbpL EL 0%toi iii poi reit toi 1.00E-03 1.20E-03 1.40E-03 1.60E-03 1.80E-03 2.00E-03 MMcf/d SRNR RRO og pe Hal TAS mt Ale BIR a nga TER toe ceea Sg AUR gahe aOL>CREE I He DRE OIE ARR ERLE TN Ae URERE He AOCAO TORI,SRE PLA AE aI I ETON RL'Page-13 Z 'Draft Work Product.-Not for.Publication eee "Sep 22,2009 | F thin st cs <i eae .See Components of the Natural Gas Decision Tree 1.0 eo Gas Storage (Supply-Management Node) -Data Sources:B&V research;ANGDA-SAIC report (2007) -Assumptions: All storage occurs in depleted oil reservoirs (DORs) All DORs are converted to storage promptly after end of production Supply-Management Node:Gas Storage -@ 2010-2019 -#-2020-2029 +2030-2039 ProbabilityofAttainment(6,)oOx10%+|/Fit as Logistic Function}: 0%; -100 200 MMecf/d _..Draft Work Product:-Not for Publication._Sep 22,2009; steers TeeIFFERENCE® F"Page-15.ssSSs«é@D raft.Work Product -Notfor Publication._ Components of the Natural Gas Decision Tree 1.0 o Regional Pipeline System (Supply-Management Node) -Data Sources:Interview with ENSTAR -Assumptions: Existing system can accommodate up to an additional 500 MMcf/d of new supplies Not a bottleneck for receipt and delivery to Railbelt gas users Supply-Management Node:Regional Pipeline System -e All Decades (2010-2059) 100%a a a :w 90%fw o-on onion Fibs Logistio Function)-)-i--)$.Cc a ee i=80%4 weeecheeee feeveebeneedfenanedenncapeccedeecendenccepeecedteeecedececcfeoeecbensedenncefeneecbeewedeaecafecccepee en @enne je---4€TOY qoreeceeceedecceebeceehececdeceecbeceedheeeedeeeecbeeedenedcebeeeebesedieeeedceedednehaQeeeeteceey60%4ev-decbecbee debe dhebbbebeedreteQuecteed- OS 5O%Fore bere beet teed ebb db LL eb eee deeeee beceecbereediercebessetben Qeeeeteneey 40%Forder bebe bebe ee te rbd bee beccebeeecficeed eee @hcetfece 1Q&30%feoeebeoe-frorecbecee sheen oo oo focsecbeeenceecerdeeeee foerecbeneedeneee oo oo oo Sone oo &DOU pore tececdeccecbeceedheccedeceecbeceedeesedececbeceedteredensesbessebestdeeeebeseesbereedienebaseesbae@ecrnteceedO%tareebewen beeen eb enee been dee chee d eed ecb ed ede ecee dened ened eee bene bene dee deeb ee Qeecebeeed 0%T T T T rT, -100 200 300 400 500 MMct/daoaRAREM52ent,BOILER TE ANEINGRIPE SG aaa SATE AOE ALT SS Es RTS,RELA NE Me ATM PE"Sep 22,2009| Be 'ee wn ME re ae et ca at More 2 Py ee oy any 63 Decision Tree 1.0 Simulation Results LACK &:VEATCH BaPreliminaryOutputfromtheNaturalGasDecisionTree1.0 Decade:2010-2019 -@-2010-2019 (20090921A) 1999-2008 Demand (R-C-I-EP) -O -2010-2019 (20090921B) 1 ProbabilityofAttainment00% 90% 80%7+- 70% 60% 50% 40% 30% 20% 10% 0% SatRtg ON-Page -16 _ sores |Of a|ao reer 234.6952,50%besseee benef418.2288,50%Of PNA |pO rn ne na Gn a nn4aeekwseeeOoSEPT,Re.Ce enn EEE See Senne See i]q Fit as Logistic Function .- weewebweeete een de eee eee ee- 100 200 300 400 Average for the Decade (MMcf/d) Latin Hypercube samplingt+) ©5,000 iterations per simulation 3 About 2-5 minutes run time per simulation Decade 2010-2019 sim No.20090921A |200909216 Cook Inlet °° Production 100%30% Bullet Line 0%25% spur Line 0%0% LNG 0%25% Gas fram 0%0%Coal Storage 0%0% Regional Nea NiAPipeline a er ee a tee ARONAFRYAgree MTEARENN TO,"Draft "Work.Product "="Notfor Publication BEERSAPT,EE APA ORO EREE TSTINKSep22,2009| ascites MinetiiscdtattiatiatidnnssiseainoPreliminaryOutputfrom the Natural Gas Decision Tree 1.0 Decade:2020-2029 Decision Tree 1.0 Simulation Results -@®-2020-2029 (20090921C)-1999-2008 Demand (R-C-I-EP)Decade 2020-2029 - ©&-2020-2029 (20090921D)Sim No.|20090921C¢|20090921D 100%a SS Cook Inlet 90%di broraee decree boarcenbereeed eo Sbebed Fit as Logistic Function --Production 80%foPegae ek Ss Se as "GR ST Bullet Line 50%50% (aaa Saas San Same moni ea Sa SSaaeeshJonoeae]fos [os40%5 ':: 30%1 ':''::'':'.'H H 'Coal a 0%ee fo)ew fo 10%4 op Sostine N/A |NSA -100 200 300 400 Average for the Decade (MMcf/d) o Latin Hypercube sampling o 5,000 iterations per simulation e About 2-5 minutes run time per simulation en 50%0% Spur Line 0%0%ProbabilityofAttainmentBER Reta EB aoe Sa nb ROSE ed a at ipeaetds A EA be aLhe AMPS mae PORE Ta eR goenant RBSHE SOME IRANa NS He SBIR Eomg eo RT SiN den crane|Page-17.-s-s-<Csé'i EL Work Product -Not.forPublication __Sep 22,2009; _GUOMGNTOROORTARS Preliminary Output from the Natural Gas Decision Tree 1.0 Decade:2030-2039 Decision Tree 1.0 Simulation Results -@ 2030-2039 (20090921E)--1999-2008 Demand (R-C-I-EP)}[Decade 2030-2039 -O -2030-2039 (20090921F)Sim No.20090921E |20090921F 100%ro a jE EE ES SS SO Cook Inlet 90%pore-Ean Souigsher Fit as Logistic Function --Production BO%Pe ccf feed ee NETOise fcrecenfenceieee encanta fatndictefce Bullet Line 50%|0% ee DO booed bean | 60%fonnecfeeendececfeceecbeceedi need ene NMG eeeefeceedheceebeceedeeee-POk=17,1588,50%60%footerthom ggg G52,60%NTE fe 50%Olen -_a LNG 2o%2% for poo epee ieeePeel [Ses fom Fon ;2 ||::|||::|:} storage 0%0% 25%0% ee eeenee eeeee Spur Line 0%75%ProbabilityofAttainmenta ae ee ee oe 10%ofenvedieneefenne donee feneedeceeboneeefeeediecnd eenfenen ieee ghee fceeedeceedaneeefeneedsneebene : fii Gd i dt i id dG EE PSE ET EE tT |Regional NfA NIA0%T T T T Pipeline { -100 200 300 400 500 Average for the Decade (MMcf/d) 0 Latin Hypercube sampling eo 5,000 iterations per simulation e About 2-5 minutes run time per simulation ORRHerHAIR:Tet Re A RECT RRR PR re OR Tite Seca nein aT AYweeQUTNR|to Ron AR,TRS eae mee Pr 8 EANa TD Orme "Sep22,2009oukeSucciaeihiekne ,DR see ee ees vaoo eae PME Se page ee I .&;VEATCH=owstnayeau7iesNaturalGasDecisionTree1.0:Mainn Takeaways Version 1.0 is only the initial step -much more work remains ahead Supply forecast results are very sensitive to input assumptions aboutprobabilityfunctionsforeachsupplynode -For each node,the probability function could experience major changes from one decade to the next -Careful review and selection is essential -reinforces importance of gas- specialist workshops to share data and vet assumptions Populating a 50-year annualized forecast matrix for each is a time- consuming computational activity -Example: (5 decades x 7 nodes)sims x 5 min/sim =175 min =2.9 hours Supply forecast is just the first part -price forecast comes later PEE OA pegs OO et ie ip Sate,Sart eee RR talheBAMIEHESeooHSiAOEGARGNEEan 'eer Page-19.oo "Draft.'Work.'Product._-"Not,for.'Publication.oe oo Sep 22,2009 :