The HomeBOTS System and Field test A Multi-Commodi

TheHOMEBOTSSystemandFieldTest:AMulti-CommodityMarketforPredictivePowerLoadManagementFredrikYgge†HansAkkermans§ArneAndersson‡MarkoKrejic¶ErikBoertjes§†EnerSearchABChalmersTeknikparkS-41288Gothenburg,[e-mail]˜ygge§FreeUniversityAmsterdamVUADepartmentofInformationManagement&SoftwareEngineering(IMSE)DeBoelelaan1081a,NL-1081HVAmsterdam,TheNetherlands{boertjes,hansa}@[e-mail]˜boertjes‡UppsalaUniversityComputingScienceDepartmentBox325S-75105Uppsala,[e-mail]˜arnea†UniversityofKarlskrona/RonnebyHK-RDepartmentofComputerScience(IDE)S-37225Ronneby,[e-mail]’99.Topics:Businessagentsforenergyservices;Negotiation;Priceformation;Toolsandtechniquesforprocesscontrol.Correspondingauthor:ErikBoertjes,FreeUniversityAmsterdam,ComputerScience/IMSEDepartment,DeBoelelaan1081a,NL-1081HVAmsterdam,TheNetherlands.E-mail:[e-mail]/[e-mail];Fax:[phone];Tel:[phone].Abstract.WepresentasystemcalledHOMEBOTSforagent-basedenergymanagementser-vices,realizedbynetworked‘smart’industrialandhouseholdequipmentcommunicatingoverthepowerlineandothermedia.Asaconsequenceofthederegulationoftheelectricitymarketsinmanycountries,energyutilitieshavestartedtopayhighinterestinofferingvalue-addedenergycustomerservicesratherthanmerelysellingelectricity(kWh).Wediscussanumberofimportanttechnicalandbusinessissuesinlaunchingsuchservices,anddescribesomeadvancedsolutions.First,wepresentanewcomputationalmarkettheory,implementedintheHOMEBOTSsystem.Itshowshowlargenumbersofelectricalloadscanbeautomaticallymanagedbyautonomousagents,thatcommunicateandnegotiateinanelectronicmulti-commoditymarketleadingtoopti-maluseofelectricalpower.Theadvantagesofthisagent-basedapproachcomparedtotraditionalmethodsforpowerloadmanagementaredescribed.Second,wedemonstratethroughsimulatedbusinessscenariosthatsignificantenergycostsavingscanthusbeachieved.Third,ourapproachhasbeentestedinafieldexperimentinanenergydistributionareaintheSouth-EastofSweden.Theperformedfieldtestsshowthatthereal-timerequirementsforagentcommunicationoverthepowerlineinenergyservicesarewellmetinrealisticapplicationsettings.1Agent-BasedEnergyManagementTheelectricitymarketinmanycountrieshasrecentlybeenderegulatedoriscurrentlyunderderegulation.Atthesametimewewitnessarapidtechnologicaldevelopment,particularlyintheareaofaccessbetweenglobalnetworks(suchastheInternet)andin-dividualhouseholds(e.g.throughcableTVoreventheelectricalpowergrid),andintheareaofincorporationofmicrocomputersinhomeappliances.Thus,animportantquestionforpowerutilitiesishowthisnewabilitytocommunicatewithlargenumbersofelectricalequipment(loads)canincreasethecompetitivenessoftheutility.Value-addedservices.Aninterestingobservationinthisnewsettingisthatutilitiescanoffernewvalue-addedservices,suchasprovidingacomfortableindoortemperatureinapublicbuildingorprovidingwaystosavecostsbyenergymanagement.Intheformercase,theutilitycouldhavesomecontractwiththecustomerwhichdefinestheeconomictermsfordifferentlevelsofcomfort(temperature).Inthelattercase,onecouldhavecontractsbetweenutilitiesandcustomerswiththeaimtoreducetheconsumptionatthecustomersideatcertain(peak)hours.Namely,energydemandvariesdependingonthenatureofthecustomerpremises,anditshowsbigandcostlyfluctuationsovertime.Onepossibilitytoincreasetheefficiencyoftheenergysystemisthereforetotryandmanagecustomerequipmenttoreducethetemporalfluctuationsindemand.Thefundamentalproblemfortheutilityforthesetwoexamplesisthesame:Whenshouldtherespectiveloadsbeatwhatconsumptionlevel(whileconsideringthecomfortconstraintsofthepub-licbuildingandthetimeconstraintsgivenbythecontracts)suchthatcostsareminimized?Conventionalpowerloadmanagement.Thisiscalledpowerloadmanagement[1,2,3].Thegoalofloadmanagementistomovedemandfromexpensivehourstocheaperhours.Thisreducescosts,curtailsenergysystemover-orunder-capacity,andenhancestheutilizationdegreeofinvestmentsinexistingenergynetworkassets.Energyloadmanagementisalreadyanoldideausedonalimitedscalewithlargecustomers.Whenbigpeaksoccur,utilitiesmaybeallowedtoshutdowncertaincustomerprocessesinreturnforacertainfinancialcompensationtothecustomer(directloadmanagement).Or,suchactionsmaybeundertakenbycustomersthemselves(indirectloadmanagement),oftentriggeredbycontractualpenaltiesiftheenergyconsumptionexceedsspecifiedceil-ings.However,thecurrentapproachestoloadmanagementarerestrictedinscaleandscope,astheyessentiallydependonone-wayhumandecisionmakingandcontrolregard-ingasmallnumberofenergy-consumingdevicesandprocesses.DuetotheexpandingcapabilitiesofInformationandCommunicationTechnology,thecurrentchallengeistoautomaticallymanageextremelylargenumbersofloadssimulta-neously.Inparticular,itshouldbecomepossibletohandleallelectricalloadsbeyondthesecondarysubstationsonthe230Vlowvoltagegrid.InEurope,atypical‘cell’beyondsuchsubstationscoversabout250households,involvingontheorderofthousandsofrelevantloads.Similarnumbersholdforofficeandpl