Page - 214 - in Emerging Technologies for Electric and Hybrid Vehicles

Energies 2016,9, 594 make predictions (namely, a power value for each interval, for each house) and requires a lot of communicationduetoexchangingpowerprofiles. Whennocoordinationcontroller isused(NOCOORD), theresultingprofile isalreadyratherflat (Figure4)andthevoltagesarewellwithin legalbounds,and,at theendof thecharging intervals, this approach isonparwithPS (Figure??). Byusingthecoordinationcontroller (COORD), theprofile is furtherflattenedandthevoltagesaresimilar to thenocoordinationcontroller for this specificcase. Table3gives theexact losses,voltagebounds,maximumpeakloadandthehighestoverloading of a cablewithin the grid. Note that several scenarios share themaximumcable load of 57.41%, andthereasonfor this is that thiscable loadoccursbefore theEVsarriveandcannotbe influenced. Thetableshowsthat thedevelopedapproachesperformsimilar toprofilesteering,while, forprofile steering,knowledgeabout the futurewasused(in therealworld,profilesteeringhas topredict these values,whereas, inourexperiments,weusedtheactualvalues). 7.Conclusions Existingdemandsidemanagementapproacheseitherdonotplanahead,andtherebyoftenmay notdeploytheflexibilityofsmartapplianceswhenit isneededthemost,or theydomakeaplanbut thisplanning isbasedonoftenhardtopredict (inaccurate)householdpowerconsumption. Toplantheapplianceswithinahouse,weproposeusingonlineplanning.Asaproofofconcept, we presented an online electric vehicle planning algorithm that only requires a prediction of the fill level characteristicandapredictionof load in thenext time intervalasan input. Thealgorithm distributes theprediction error evenlyover all charging intervals. Thismakes the algorithmvery robustagainst incorrectpredictions, especially if thepredictionsarehigher thantheactual realization. Furthermore,wepresentedaboundthatquantifies thesensitivityofourapproachtopredictionerrors. Weextendedthehousecontrolmechanismbyaneighborhoodcontrolmechanism,which initially asks thehouses tomakeaflatprofile.Whenthe loadexceedsacertain threshold, this coordination controller requests lesschargingfromhouses for thenext timeperiod insuchawaythat thehouse profiles remainasflat aspossible. Thismethodonly requires apredictionof thenumberof active charging intervalswithinahouse tomaketherequiredtrade-off. Both thehouseandneighborhoodapproachesrequireonly fewpredictions,andweshowedthat thesepredictionsareeasy toobtain. In theevaluation,westudiedthecombinationofpredictionswith ourapproach,anddemonstratedthatitworkseffectivelyforpeak-shaving.Theevaluationfurthermore showsthat thisapproach isveryrobustagainstpredictionerrors, andperformsadequatelyevenwhen thepredictionsarevery imprecise. Comparedtoanaiveapproach, it leads to lower transportation losses,keeps thevoltagewithinrequiredboundsandkeepscable loads low. Furthermore, it isonpar with thestate-of-the-art research,which—incontrast toourwork—requirespredictionsofflexibility andpredictionsofa loadprofile (24hoursahead) foreachhouse. In futurework,weaimtoextend theonlineplanningwithinahouse to copewithappliances other thanelectricvehicles. Acknowledgments:This researchwasconductedwithin theDREAMproject supportedbySTW(#11842)and the e-balance project that has received funding from the EuropeanUnion Seventh Framework Programme (FP7/2007-2013)undergrantagreementn◦ [609132]. AuthorContributions:Theconcepts, ideasandtheorydescribed in thisarticleare jointworkbybothauthors. ThesimulationswereconductedbyMarcoE.T.Gerards. Conflictsof Interest:Theauthorsdeclarenoconflictof interest. 214