Assessment of the regionalised demand response potential in Germany using an open source tool and dataset
Wilko Heitkoetter, Bruno U. Schyska, Danielle Schmidt, Wided Medjroubi, Thomas Vogt, Carsten Agert
AAssessment of the regionalised demand response potential in Germany using an opensource tool and dataset
Wilko Heitkoetter a , Bruno U. Schyska a , Danielle Schmidt a , Wided Medjroubi a , Thomas Vogt a , Carsten Agert a a DLR Institute of Networked Energy Systems, Carl-von-Ossietzky-Str. 15, Oldenburg, Germany, [email protected]
Abstract
With the expansion of renewable energies in Germany, imminent grid congestion events occur more often. One approach for avoid-ing curtailment of renewable energies is to cover excess feed-in by demand response. As curtailment is often a local phenomenon,in this work we determine the regional demand response potential for the 401 German administrative districts. The load regionali-sation is based on weighting factors derived from population and employment statistics, locations of industrial facilities, etc. Usingperiodic and temperature-dependent load profiles and technology specific parameters, e.g., the time frame of management, loadshifting potentials were determined with a temporal resolution of 15 minutes. Our analysis yields that power-to-heat technologiesprovide the highest potentials, followed by residential appliances, commercial and industrial loads. For the considered 2030 sce-nario, power-to-gas and e-mobility also contribute a significant potential. The cumulated load increase potential of all technologiesranges from 5 − MW per administrative district. The median value is 25 MW , which would su ffi ce to avoid the curtailmentof 8 classical wind turbines. Further, we calculated load shifting cost-potential curves for each district. Industrial processes andpower-to-heat in district heating have the lowest load shifting investment cost, due to the largest installed capacities per facility. Wedistinguished between di ff erent size classes of the installed capacity of heat pumps, yielding 23% lower average investment costfor heat pump flexibilisation in the city of Berlin compared to a rural district. The variable costs of most considered load shiftingtechnologies remain under the average compensation costs for curtailment of renewable energies of 110 e / MWh . As all results andthe developed code are published under open source licenses, they can be integrated into energy system models, which is simplifiedby the applied storage equivalent demand response formulation.
Keywords: demand response, load shifting, regionalisation, cost-potential curves, open data, open source
1. Introduction
In pursuit of reducing global CO emissions and mitigat-ing climate change, renewable energy sources are consid-ered a main instrument [1], constituting 33% of global gen-eration capacity in 2018 [2]. However, the intermittent andnon-dispatchable feed-in of variable renewable energy sources(VRE) [3] requires balancing technologies, such as dispatch-able generators, energy storage or transmission line reinforce-ment [4].Another balancing option that plays a minor role cur-rently [5], but may gain importance with increasing shares ofrenewable energies, is demand response (DR) [6]. Demandresponse utilises available elasticity of consumer demand andcomprises two classes [7]: Load shedding applies to loads be-ing reduced, but for which cannot be compensated for at anothertime [8]. Load shifting is associated with loads being shiftedto an earlier or later time, e.g., from a period with low VREfeed-in to a period with high VRE feed-in [4]. In this paper wefocus on the application of demand response for avoiding VREcurtailment in times where feed-in exceeds demand or grid ca-pacity. As this service can only be provided by load shifting,load shedding is disregarded in our analysis.Table 1 shows selected studies that provide spatially and tem-porally resolved data for the load shifting potential in Germany. The studies [4, 8, 9, 10, 11, 12, 13, 14] analyse the potential forGermany, while [15] focuses on multiple countries in NorthernEurope and the geographical scope of [7, 16] is entire Europe.Most of the mentioned studies determine aggregated load shift-ing potential values on country level (NUTS-0 [17]). Such aspatial resolution is su ffi cient, if the influence of load shiftingon national power markets shall be investigated. For example,Klobasa [10] addresses decreased planning horizons in conven-tional power plant dispatch due to increased VRE penetrationand assesses how the overall power system e ffi ciency can beincreased by the application of load shifting.In the case that load shifting for the avoidance of VRE cur-tailment shall be assessed, a higher spatial resolution is re-quired. VRE curtailment is mostly caused by surplus feed-inand limited grid transport capacity. As these factors depend onthe installed capacity and energy demand in a specific region,as well as on the power flow from or to other regions [18], VREcurtailment is often a local phenomenon. The studies [7, 13]account for such a higher spatial resolution and provide loadshifting potential values on administrative district level (NUTS-3 [17]). We henceforth denote the process of allocating data toregions of a territory as “regionalisation”.The selected studies also di ff er with respect to the tempo-ral resolution of the results. In [8, 10, 12, 13, 15] no temporal1 a r X i v : . [ phy s i c s . s o c - ph ] S e p tudy Spatial scope / spatial resolution / temp. resolution Electricity sector:residential / commercial / industrial Sector coupling:power-to-heat / e-mobility / power-to-gas Potential restrictions:technical / socio-technical / economic Openness andreproducibilityStadler [9] GER / NUTS-0 / h (cid:88) / (cid:88) / (cid:88) (cid:88) / – / – (cid:88) / ( (cid:88) ) / ( (cid:88) ) grey boxKlobasa [10] GER / NUTS-0 / – (cid:88) / (cid:88) / (cid:88) (cid:88) / – / – (cid:88) / ( (cid:88) ) / (cid:88) grey boxPaulus et al. [8] GER / NUTS-0 / – – / – / (cid:88) (cid:88) / – / – (cid:88) / – / (cid:88) grey boxElberg et al. [11] GER / NUTS-0 / h (cid:88) / (cid:88) / (cid:88) (cid:88) / (cid:88) / – (cid:88) / ( (cid:88) ) / (cid:88) grey boxApel et al. [12] GER / NUTS-0 / – (cid:88) / (cid:88) / (cid:88) (cid:88) / – / – (cid:88) / ( (cid:88) ) / ( (cid:88) ) grey boxGils [7] EUR / NUTS-3 / h (cid:88) / (cid:88) / (cid:88) (cid:88) / (cid:88) / – (cid:88) / – / (cid:88) grey boxvon Bremen et al. [16] EUR / NUTS-0 / h (cid:88) / (cid:88) / (cid:88) (cid:88) / (cid:88) / – (cid:88) / – / – grey boxS¨oder et al. [15] N.EUR / NUTS-0 / – (cid:88) / (cid:88) / (cid:88) (cid:88) / – / – (cid:88) / – / – grey boxPellinger et al. [13] GER / NUTS-3 / – (cid:88) / (cid:88) / (cid:88) (cid:88) / (cid:88) / (cid:88) (cid:88) / ( (cid:88) ) / (cid:88) grey boxM¨uller et al. [4] GER / NUTS-0 / h (cid:88) / (cid:88) / (cid:88) (cid:88) / – / – (cid:88) / – / – grey boxSteurer [14] GER / NUTS-1 / h (cid:88) / (cid:88) / (cid:88) (cid:88) / – / – (cid:88) / (cid:88) / (cid:88) grey boxPresent study GER / NUTS-3 / (cid:88) / (cid:88) / (cid:88) (cid:88) / (cid:88) / (cid:88) (cid:88) / (cid:88) / (cid:88) white boxAnnotation: - = not considered; (cid:88) = considered; ( (cid:88) ) = limited consideration;white box = applied equations, source code and data are publicly available; grey box = significant part of applied equations, source code or data remains undisclosed Table 1: Overview of selected studies taking into account the spatially and temporally resolved load shifting potential in Germany. resolution of the data is taken into account and instead mini-mum and maximum values for the load shifting potential areprovided. The resulting potential data in [11, 14, 19] are givenwith a temporal resolution of 1 hour.Processes that are suitable for demand response typicallyprovide thermal inertia, a physical storage or demand flexibil-ity [7]. In the residential sector, white goods such as washingand drying machines, or fridges and freezers, are considered assuitable. Common DR applications in the commercial sectorare ventilation and cooling appliances. In the industrial sec-tor there are processes with a physical storage that are particu-larly fitted for load shifting, e.g., cement mills or wood pulpers.Cross-sectional technologies are associated with multiple in-dustry branches, e.g., compressors for pressurised air. Loadshedding processes regularly run at their full installed capacityand can thus only be switched o ff , not shifted, e.g. metal pro-duction. Most of the considered studies take into account all ofthe DR application categories, except [8], which only focuseson the industrial sector.In the course of decarbonisation the power sector will in-creasingly be coupled with the heating, transport and gas sec-tor [20]. While power-to-heat (PtH) technologies are consid-ered in all investigated studies, the e-mobility sector is in-cluded on NUTS-0 level only by Elberg et al. [11] and spa-tially resolved results on NUTS-3 level are only providedby [7] and [13]. The power-to-gas sector is only consideredby Pellinger et al. [13].The theoretical load shifting potential is further limited bytechnical restrictions, e.g. maximum load increase and decreasefactors [14]. Considering additional acceptance and organisa-tional constraints yields the socio-technical potential. The eco-nomic potential is calculated by taking cost and revenue param-eters into account. While all analysed studies regard technicalrestrictions, some of the studies do not consider socio-technicaland economic restrictions, or only consider these in a limitedmanner, e.g., by not providing explicit restriction factors.In the present study all mentioned technologies and poten-tial restrictions are considered. Additionally, we include cen-tralised PtH in district heating and in the industry, which wasnot regarded in the considered studies. For modelling residen- tial PtH more than 700 building types are used to account forthe regional di ff erences in the residential building stock. Theload shifting potential results are provided with a high spatialresolution on NUTS-3 level and a temporal resolution of 15minutes. We determine the economic potential in the form ofregional cost-potential curves for each administrative district.As a case study, we investigate the influence of heat pump sizeclasses on the regional cost-potential curves. In contrast to otherstudies, all result data and developed source code are publishedopen source. This supports external model evaluation by otherresearchers and avoids duplication in data collection and codeimplementation [21]. In particular, this paper will examine thefollowing research questions: • What are the load shifting potentials on NUTS-3 level inGermany? • How does the potential di ff er between rural and urban dis-tricts? • How are the regional load shifting cost-potential curvescharacterised? • How will the load shifting potential develop in future?
2. Methods
The methodology of this work comprises the NUTS-3 loadregionalisation and the determination of the load shifting poten-tials, as depicted in Figure 1. We implemented the computationin the open source python tool dsmlib, which can be obtainedfrom the supplementary material, together with the used inputdata and obtained result data .In this section we first describe the equations of the load shift-ing model (Section 2.1). Next, the determination of the inputdata for the model is presented, starting with the regionalisationof the energy demand and maximum capacity for all consid-ered technologies (Section 2.2). In Section 2.3 the calculation For data and source code refer to the supplementary material of this articleat: https: // doi.org / / zenodo.3988921 OAD
SHIFTING
MODELNUTS-3
Scheduled load L(t)
NUTS-3 Maximum capacity Λ Time frame ofmanagement Δ tLoad increase / decreaseshares s inc / s dec NUTS-3Load increase / decrease potentials P max / P min NUTS-3Energy preponing / postponing potentialsE max / E min
NUTS-3REGIONALISATIONNUTS-0 energy demandRegionalisationweightsLoad profilesFlexible share s flex
Utilisation rate s util
COSTASSIGNMENT Load flexibilisationcosts c inv , c fix , c var
NUTS-3 cost-potential curves
Figure 1: Overview of the process for deriving the regionalised load shifting potentials and cost-potential cuvces, implemented in the open source tool dsmlib. of the scheduled load is introduced. Subsequently, the deter-mination of the remaining input parameters is explained: thetime frame of management (Section 2.4), the load increase anddecrease shares (Section 2.5), the flexible share (Section 2.6)and the load shifting costs (Section 2.7). While we refer to the2018 status quo scenario in the previous sections, in Section 2.8the parameters for modelling the 2030 future scenario are pre-sented.
In order to ease the integration into energy system models,in this work load shifting is modelled as an energy storage-equivalent operation. The model is an enhancement of a for-mulation that was developed at the DLR Institute of NetworkedEnergy Systems in [22]. The following input parameters areused by the model for each regarded category, c , of load shift-ing technologies:1. Scheduled load L c ( t ): Load time series for a given appli-cation, without any load shifting modifications.2. Time frame of management ∆ t c : Maximum duration bywhich loads can be postponed or preponed.3. Maximum capacity Λ c : Installed capacity of a given ap-plication.4. Load increase and decrease limits, s cinc and s cdec : Share ofthe maximum capacity up to which the load can be in-creased or decreased.The load shifting process of the scheduled load L c ( t ) results in anew time series, the realised load R c ( t ). In terms of load shiftingas an energy storage-equivalent operation, the charging rate ofthe storage can be defined as: P c ( t ) = R c ( t ) − L c ( t ) , (1)where the storage is charged for P ( t ) > P ( t ) <
0. Integrating the charging rate over time yields the The load increase and decrease parameters ( s inc / s dec ) were added in thiswork. filling level E ( t ) of the storage, E c ( t ) = (cid:90) t P c ( t (cid:48) ) dt (cid:48) . (2)Using this terminology storage-equivalent bu ff ers can be de-fined as: E cmax ( t ) = (cid:90) t +∆ tt L c ( t (cid:48) ) dt (cid:48) , (3) E cmin ( t ) = − (cid:90) tt − ∆ t L c ( t (cid:48) ) dt (cid:48) , (4) P cmax ( t ) = Λ c · s cinc − L c ( t ) , (5) P cmin ( t ) = − ( L c ( t ) − Λ c · s cdec ) . (6)These bu ff ers serve as boundary conditions for the charging rateand filling level of the storage: E cmax ( t ) ≤ E c ( t ) ≤ E cmin ( t ) ∀ t , (7) P cmax ( t ) ≤ P c ( t ) ≤ P cmin ( t ) ∀ t . (8)The load increase and decrease bu ff ers, P max and P min , and theenergy preponing and postponing bu ff ers, E max and E min , areschematically illustrated in Figure 2 and 3. As these bu ff erscan be determined before the load shifting dispatch optimisa-tion, the presented model allows for a computationally e ffi cientintegration of load shifting into energy system models.The cumulated time series of all technologies for the sched-uled and realised load, as well as the bu ff ers, are obtained bysumming over all categories c . In the following sections wedescribe the determination of the input data for the load shift-ing model and the utilised input parameters are summarised inTable 3. The maximum capacity Λ ci per NUTS-3 district i of each loadshifting technology is determined by using the NUTS-3 annualenergy demand E ci and the average annual utilisation rate s cutil : Λ ci = E ci · s cf lex h · s cutil (9)3 omenclature Selected abbreviations
AC Air conditioningBDEW German Association of Energy and Water IndustriesCOP Coe ffi cient of performanceCTS Commercial, trade and servicesDHW Domestic hot waterDR Demand responseGVA gross value addedNUTS Nomenclature des Unit´es territoriales statistiquesNUTS-0 Country levelNUTS-3 Administrative district level (401 districts in Germany)PtG Power-to-GasPtH Power-to-HeatVRE Variable renewable energy sources Selected Variables, Parameters and Indices c Load shifting technology index – i Administrative district index – ∆ t Time frame of management hL Scheduled load MW Λ Maximum capacity
MWs dec / inc Load decrease / increase limit – P min / max Maximum load decrease / increase MWE min / max Maximum energy postponing / preponing MWhs flex
Flexible share – s util Utilisation rate – c inc Specific investment costs e / MWc fix
Specific annual fixed costs e / MW / ac var Specific variable costs e / MWhx
Relative change of installed capacity until 2030 – P Installed capacity in 2030 GW Figure 2: Schematic illustration of the load increase and decrease bu ff ers, P max and P min .Figure 3: Schematic illustration of the energy preponing and postponingbu ff ers, E max and E min . The flexible share parameter, s cf lex , accounts for socio-technicalrestrictions (see Section 2.6). In the next subsections theNUTS-3 annual energy demand is derived for the di ff erent de-mand sectors and technologies. Table 2 gives an overview ofthe numeric values of the utilised parameters for the energy de-mand regionalisation. In the residential sector washing and drying machines, aswell as fridges and freezers are suitable for load shifting [10].Therefore, we multiplied the annual electricity demand of theGerman residential sector, ¯ E res , with the share of washingand drying machines, x res , wd , and fridges and freezers, x res , f f .Hence, the NUTS-3 annual energy demand, E res , wd / f fi , was de-termined via: E res , wd / f fi = x resi · x res , wd / f f · ¯ E res , (10)where x resi is the share of each administrative district in theGerman residential electricity demand. The electricity demandper resident rises with higher income and fewer members perhousehold [13]. Thus x resi was computed by: x resi = ( I i sc (cid:88) sc = ω sc N i , sc ) / ¯ I . (11)4 ector Symbol Value Unit Description SourceResidential ¯ E res
129 TWh Annual electricity demand of the German residential sector [25] x res , wd x res , f f E cts
147 TWh Annual electricity demand of the German CTS sector [25] x cts , ve x cts , ac x cts , co E ind
239 TWh Annual electricity demand of the German industry sector [26] x ind , ve x ind , co < COP hp > ffi cient of performance of heat pumps [13] < COP rs > ffi cient of performance of resistive heaters [23]¯ Q cts
216 TWh Annual heat demand of the German CTS sector [25] x cts , hp x cts , rs e ev n ev Table 2: Input parameters for the energy demand regionalisation for 2018. Additional input parameters are given in previous works of the authors on the regionali-sation of the residential heat demand [23] and industrial energy demand [24].
Therein I i is the average income per inhabitant per district and ¯ I the cumulated income of all residents of Germany [31]. In orderto account for the increasing electricity demand per householdmember of households with fewer members [32], we consid-ered six household size classes m sc ∈ { , , , , , } . Next, theweighting factor ω sc was defined by, ω sc = . · m sc + . , (12)and multiplied with the number of households per householdclass in each administrative district, N i , sc . In the commercial, trade and services (CTS) sector cross-sectional technologies that are used in multiple CTS branchesprovide the highest load shifting potential [10]. Out of thecross-sectional technologies, ventilation and air conditioningappliances, as well as cooling processes are considered as suit-able for load shifting [10]. We multiplied the overall GermanCTS electricity demand, ¯ E cts , with the share that is needed forventilation, x cts , ve , for air conditioning, x cts , ac , and for cool-ing, x cts , co [27]. To determine the NUTS-3 distribution of theregarded processes energy demand, E cts , ve / ac / coi , the share ofthe CTS employment of the respective administrative district, x ctsi [33], in the overall German CTS employment was used: E cts , ve / ac / coi = ¯ E cts · x cts , ve / ac / co · x ctsi (13) As like as in the CTS sector, also in the industrial sector, non-process relevant ventilation appliances and cooling processesin the food industry are considered as suitable for load shift-ing [10]. The methodology for determining these load shiftingpotentials is equivalent to Eq. 13, but instead uses the respectiveenergy demand and demand shares in the industry, ¯ E ind , x ind , ve , x ind , co and x indi [34, 35]. For more information refer to the supplementary material.
Furthermore, the following energy intensive industrial pro-cesses are suitable for load shifting [7]: cement milling, me-chanical wood pulping, paper production, recycled paper pulp-ing, and air separation. In [24] the authors of this study showthat it leads to significant errors, when the energy intensive pro-cesses are regionalised using statistical NUTS-3 employment orGVA data. Instead, we identified individual industrial plants, aswell as their locations, and production capacities, c pl , by usingregisters of German national industry associations. The produc-tion capacities were multiplied with the specific energy demand e pl and the utilisation factor, s util , pl . All plants per administrativedistrict i were summed to determine the annual energy demandof the energy intensive processes, E ind , inti = (cid:88) pl c pl · e pl · s util , pl . (14)The plant specific regionalisation methodology is described indetail in [24] and the results are published as an open dataset. In order to assess the role of sector coupling technologiesfor load shifting, we clustered the electric heating technologiesfrom the residential, CTS and industry sector as one separatepower-to-heat sector.
Residential power-to-heat
The regionalisation of the residential power-to-heat load isbased on an open dataset developed in a previous work of theauthors [23]. Using a special evaluation of census enumera-tion data, 729 residential building categories b were defined.An area specific annual heat demand q (cid:48)(cid:48) b was assigned to eachbuilding category, depending on the year of construction of thebuildings, the type of building, number of flats per building,floor area, heating type and number of residents per building.To yield the absolute annual heat demand, the area specific heat The open dataset is available at: https: // doi.org / / zenodo.3613766 The open dataset is available at: https: // doi.org / / zenodo.2650200 A b of thebuildings in the respective category, as well as the number ofbuildings per category per administrative district n b . To calcu-late the electrically-covered heat demand, the share of buildingswas taken into account that are equipped with heat pumps, x hpb ,and resistive heating devices x rsb . The demand of all buildingcategories b was summed, yielding the overall demand per ad-ministrative district. The annual electricity demand E hp / rsi wasdetermined by dividing the heat demand by the average annualcoe ffi cient of performance < COP hp / rs > [23]: E sh , hp / rsi = < COP hp / rs > − · (cid:88) b q (cid:48)(cid:48) b · A b · n b · x hp / rsb (15)Next, the load shifting potential of electric domestic hot wa-ter (DHW) heaters was determined. An average annual demandper person q dhw was assigned and multiplied with the numberof residents n resb in the respective building category. As definedin Eq. 15 it was summed over all building categories and theheat demand was converted to electric energy demand: E dhw , hp / rsi = < COP hp / rs > − (cid:88) b q dhw · n resb · n b · x hp / rsb . (16)The installed capacities of centralised PtH facilities in districtheating grids are given in [36]. We assigned these capacities tothe respective administrative districts, in which the district heat-ing grids are located. According to [36], only resistive heatersare installed in German district heating grids, while there are nolarge-scale heat pumps. CTS power-to-heat
For the regional distribution of commercial building types,comprehensive statistical data such as for residential buildingsdoes not exist. We therefore regionalised the overall Germanannual heat demand in the CTS sector, ¯ Q cts , to the NUTS-3level, according to the share of the NUTS-3 annual gross valueadded (GVA), x gvai [37], in the total German GVA. The CTSheat demand was converted to electricity demand, E cts , hp / rsi , us-ing the demand share covered by heat pumps x cts , hp [28] andby resistive heating x cts , rs [25], as well as the respective annualaverage coe ffi cients of performance < COP hp / rs > : E cts , hp / rsi = ¯ Q cts · x gvai · x cts , hp / rs ∗ < COP hp / rs > − . (17) Industrial power-to-heat
The regionalisation of the industrial process heat demand isbased on a previous work of the authors [24] and is summarisedin this section. Amongst others, there is a significant processheat demand in the following industries: metal, minerals, min-ing, food, textile, paper, machinery and wood [24]. The annualNUTS-0 primary energy demand of the respective processes¯ E ind , pribr [24] was multiplied with the shares of the process heat x phbr [24]. Next, it was multiplied by the average conversion ef-ficiency to useful heat energy η ind , phbr [24] per industrial branch,which depends on the used primary energy carriers, e.g. coal,oil or natural gas. The NUTS-0 process heat demand was dis-tributed to the administrative districts using the share of the NUTS-3 employment in the considered branches x i , br [24] inthe total German employment in those branches. We summedover all branches and divided the result by the average coef-ficient of performance, < COP rs > , to yield the electric energydemand for covering industrial process heat in each administra-tive district, E i = < COP rs > − (cid:88) br ¯ E ind , pribr · x phbr · η ind , phbr · x i , br . (18)There were no statistical data on the number of installed heatpumps in industrial facilities available. Therefore, we assumedthat all industrial PtH is covered by resistive heaters. Next, we determined the annual energy demand of electricvehicles, E evi , in all German administrative districts. To calcu-late this, the annual electrical energy demand per electric ve-hicle, e ev , was multiplied with the number of plug-in electricvehicles in Germany n ev and the share of registered electric ve-hicles per administrative district, x evi [30], in all electric vehiclesin Germany: E evi = e ev · n ev · x evi . (19)In contrast to the other considered technologies, the maximumcapacity is not constant for e-mobility, as the share of the con-nected electric vehicles, x ev , connt , varies during the day. Thus thetemporally resolved maximum capacity Λ evt is calculated by, Λ evt = Λ ev · x ev , connt . (20)The values of x evt are provided in [38] and were derived fromthe German mobility statistics [39]. For the NUTS-3 regionalisation of power-to-gas capacities Λ hy / mei , we assigned the locations of the PtG plants listed in [40]to the respective administrative districts they are located in. Wedi ff erentiated between capacities of power-to-hydrogen plants, c hypl and power-to-methane plants, c mepl and summed the capaci-ties per district: Λ hy / mei = (cid:88) pl Λ hy / mepl . (21) The scheduled load time series, L i ( t ), for each administrativedistrict and each technology were calculated via the annual en-ergy demand, the flexible share, the share of the energy demandin each time step, x L ( t ), and the time step length, δ t : L i ( t ) = E i · s f lex · x L ( t ) δ t . (22)The values for x L ( t ) were determined using normalised dailyload profiles, as well as additional yearly load profiles for thePtH and cement milling technologies. In the following sec-tions, the applied methodology is summarised and the resultingload profiles are visualised in Figure 6. For further informationand the numerical values of the load profiles, refer to the givensources and the supplementary material of this manuscript.6 .3.1. Daily load profiles A commonly used reference for daily load profiles are thestandard load profiles of the German Association of Energy andWater Industries (BDEW) [41]. We used the BDEW standardload profile for households, H
0, for modelling the load profileof the residential appliances. The considered cross-sectionaltechnologies in the CTS sector, cooling, ventilation and AC,have high utilisation rates. Therefore the BDEW standard loadprofile G GHD , for commercial, trade andservices [43]. For PtH covering industrial process heat the loadprofiles were adopted from [7] and take into account demandshares and typical full load hours of di ff erent manufacturingbranches.Charging of electric vehicles was assumed to follow the loadprofile developed in [38]. The profile was derived from Germanmobility statistics data [39] and contains a morning and eveningcharging peak. Power-to-gas plants were treated as energy in-tensive industrial processes and the load profile was approxi-mated to be constant. Out of the considered technologies only PtH for space heat-ing and cement production are assumed to have a varyingload profile during the course of the year, because these pro-cesses are dependent on the ambient temperature. For mod-elling the yearly load profile of PtH in the residential sector, themethodology developed by the authors of this paper in [23] wasadopted. The space heating load ˙ Q b , t at each point of time t de-pends on the di ff erence between the ambient temperature T ambt and the heating limit temperature T hl , as well as on the ambienttemperature specific heat demand factor H . Both, T hlb and H b depend on the building attributes, e.g. insulation quality or floorarea. Therefore individual T hlb and H b values were assigned foreach residential building category [23], which were introducedin Section 2.2.4. The heat load for space heating, ˙ Q b , t , couldthus be calculated, as follows: T amb ≥ T hl : ˙ Q b , t = , (23) T amb < T hl : ˙ Q b , t = H b · ( T hlb − T ambt ) . (24) For T ambt we used the daily average ambient temperaturesfrom 2018, measured at the closest weather station to the re-spective administrative district, which were obtained from thewebsite of the German Meteorological Service [44]. Due to thethermal inertia of the building mass, the temperatures of pre-vious days influence the daily heat load [45]. To account forthis influence, we applied a geometric series to the temperatureinput data, considering the three previous days, as described indetail in [23].Further, ˙ Q b , t was multiplied with the number of buildings perbuilding category and administrative district, as well the shareof buildings equipped with heat pumps and resistive space heat-ing. The result was summed over all building categories anddivided by the temporally resolved coe ffi cient of performance,yielding the scheduled electric load for space heating per ad-ministrative district: P sh , hp / rsi , t = ( COP hp / rst ) − (cid:88) b ˙ Q b , t · n b · x hp / rsb (25)While the coe ffi cient of performance for resistive space heating, COP rst , was assumed to be constant, the coe ffi cient of perfor-mance of heat pumps, COP hpt , depends on the ambient temper-ature. Therefore, we adopted an approach by Ruhnau [46], whocalculates the COP time series of air sourced (ASHP), groundsourced (GSHP) and water sourced (WSHP) heat pumps basedon a quadratic regression of manufacturer data: COP hpt = . − . · ∆ T + . · ∆ T , AS HP , . − . · ∆ T + . · ∆ T , GS HP , . − . · ∆ T + . · ∆ T , WS HP . (26)Therein ∆ T comprises all possible combinations of source andsink temperatures, ∆ T sink , source = T sink − T source , (27)where source ∈ { air , ground , water } and sink ∈ { radiator heating , f loor heating } . For T source ,hourly measured air and ground temperatures, measured at theclosest weather station to the respective administrative district,were obtained from the German Meteorological Service [44].For WSHP a constant ground water temperature of 10 ◦ C throughout the year was assumed, following [46]. The sinktemperatures depend on the utilised sink type, as well as theambient temperature. For the calculation of T sink , as well asthe assumed heat transfer temperature di ff erences, refer tothe supplementary material of this paper or the descriptionof the methodology given in [46] that was adopted here.Subsequently, a weighted average of COP time series forthe di ff erent technologies was calculated using the shares ofinstalled heat pumps in Germany, 55% ASHP, 39% GSHP and6% WSHP [48, 49].As introduced in Section 2.2.4, no data were available fora detailed regionalisation of commercial building types. Wetherefore used the BDEW annual heat demand load profile [43] For more information on heat pump types, refer to [47]. . ◦ C as threshold temper-ature, below which cement mills are assumed to be shut downand their load profile is set to zero. The time frame of management parameter, ∆ t c , specifies themaximum duration by which loads can be preponed or post-poned and mainly depends on the storage capacity of the con-sidered processes [7]. For heating applications the thermal in-ertia of the building and the size of the hot water storage are thedetermining factors, for industrial processes the product stor-age capacity is decisive and for electric vehicles the installedbattery capacity. Residential washing and drying appliances arethe only considered processes that do not have a physical stor-age capacity. The time frame of management parameter speci-fies for how long these processes can be shifted without signif-icantly disturbing the user. In Table 3, the numerical values for ∆ t are provided, as well as the respective literature sources. In case that there are no technical restrictions for decreasingor increasing the load for a specific appliance, s cdec is set to zeroand s cinc is set to one. However, for some of the consideredprocesses a complete shut-down due to temporary load shiftingis not possible. In air separation plants for example this coulddamage the technical facilities or impair product quality [14].For such processes s cdec is set to a value between zero and one.For the industrial processes a revision outage of 5% of thetime of a year is assumed [14]. This results in an average loadincrease limit of s cinc = .
95. As shown in Table 3, residentialwashing and drying appliances as well as domestic hot waterheaters have very low utilisation rates. Due to usage prefer-ences [52], demand response can only lead to a limited increaseof usage at a specific point of time, yielding s cinc values below0.2 [14]. The numerical values of s cdec and s cinc for all consideredprocesses are provided in Table 3. To account for socio-technical load shifting potential restric-tions, the annual energy demand, E ci , was multiplied with theflexible share parameter, s cf lex , in Eq. 9 and Eq. 22. The follow-ing restrictions are summarised in s cf lex [14]: First, the organ-isational feasibility limits the potential, e.g., in the case that achange of working hours is necessary to allow for load shiftingin the CTS or industrial sector. Second, the social acceptance isa limiting factor, for example if load shifting a ff ects usage pref-erences of residential appliances. Third, the regulatory frame-work can hamper the implementation of load shifting. For each considered technology the s cf lex parameter values are presentedin Table 3. Load shifting costs can be divided into specific investmentcosts c inv , annual fixed costs c f ix and variable costs c var [53].The investment is made up of the costs for information andcommunication technology (ICT) components, as well as in-stalling and programming of the devices [14]. The annual fixedcosts are caused by maintenance works and the electricity con-sumption of the ICT components [14]. The variable costs re-flect compensations for losses in production outputs and com-fort [53]. In Table 3 the assumed cost parameters and the useddata sources are given.We further analysed the load shifting investment cost for dif-ferent PtH size classes in the residential sector, because thebuilding structure significantly di ff ers between rural and urbanareas [23]. As described in Section 2.2.4 we defined more than700 residential building categories and assigned a heat demandvalue to each category. The buildings were grouped accord-ing to the heating types: single-storey heating, central heatingand district heating. For the central heating technology, we dis-tinguished between three classes of installed thermal heatingcapacity: ˙ Q inst < . kW th , . kW th < ˙ Q inst < kW th and 25 kW th < ˙ Q inst [23]. To determine the installed electriccapacity, P el , the heating capacity, ˙ Q inst , was divided by theannual average coe ffi cient of performance of the electric heat-ing, < COP > . Due to the greater energy e ffi ciency [54] and thehigher number of new installations [23], we only consideredheat pumps in this detailed cost investigation and neglected re-sistive heating technologies. The investment costs for the flex-ibilisation of heat pumps is estimated at 310 e in [13]. Thisnumber is divided by the P el values of the di ff erent size classesto determine the specific investment costs, c inv . The numericalvalues for P el and c inv are presented in Table 4. To model the future trend of the load shifting potential inGermany, we took into account a scenario for 2030. While thedevelopment of the installed capacities of the di ff erent technol-ogy classes was considered, the load profiles and load shift-ing parameters ( ∆ t , s util , s dec , s inc , s f lex ), as well as the specificcosts, were assumed not to change. Due to decarbonisation andthe associated electrification of the heating and transport sector,the installed capacities of the e-Mobility, PtH and PtG technolo-gies are expected to increase. We therefore used the GermanGrid Development Plan for 2030 [50] as a reference for the fu-ture scenario, as it provides consistent predictions for the men-tioned sector coupling technologies. Within the German GridDevelopment Plan there are di ff erent scenarios considered. Inthe present paper, we adopted scenario ”B”, which assumes amoderate speed of decarbonisation and flexibilisation of the en-ergy system.The other considered technology classes in the residential,CTS and industrial sector are very specific to load shifting andthere are no detailed future projections for these technologies in8 ector Technology, c ∆ t c [ h ] s cutil [–] s cdec [–] s cinc [–] s cflex [–] x c / P c c cinv [ e / MW ] c cfix [ e / MW / a ] c cvar [ e / MWh ]Residential Washing,drying 6 [7] [14] † † [14] [14] ∗ ;[13] ∗ ;[13] [7] Cooling,freezing 2 [7] [10] [14] [14] [14] [14] ∗ ;[13] ∗ ;[13] ∗ ;[7] CTS Cooling,ventilation,AC 1 [7] [7] [14] [14] [14] ∗ ;[14] [7] [7] [7] Industry Air separation 4 [10] [14] [14] [14] [10] [14] [14] [14] [14]
Cement 4 [10] [14] [14] ¶ [14] [14] [14] [14] [14] Pulp 2 [10] [14] [14] [14] [14] [14] [14] [14] [14] Paper 3 [7] [14] [14] [14] [14] [14] [14] [14] [14] Recycled paper 3 [7] [14] [14] [14] [14] [14] [14] [14] [14] Cooling 2 [7] [7] [7] [7] [14] [7] [7] [7] [7] Ventilation 1 [7] [7] [7] [7] [14] [7] [7] [7] [7] PtH Process heat (ind) 3 ¶ ¶ ¶ ¶ ¶ ¶¶ ¶ ¶ ¶ Heat pumps (res) 3 [14] [23] [14] [14] [14] [50] ∗ ;[13] ∗ ;[13] ∗ ;[7] Resistive sh. (res) 12 [7] [23] [14] [14] [14] [23] ∗ ;[13] ∗ ;[13] ∗ ;[7] Resistive dhw. (res) 12 [7] [7] [14] [14] [12] [23] ∗ ;[13] ∗ ;[13] ∗ ;[7] PtH in district heating 12 § [23] § ¶ S [23] ‡ ‡ ¶ Heat pumps (cts) 3 [14] [23] [14] [14] [14] ¶¶ ∗ ;[7] ∗ ;[7] ∗ ;[7] PtG Power-to-methane 24 [51] [50] ¶ ¶ ∗ [50] ‡ ‡ ‡ Power-to-hydrogen 24 §§ [50] ¶ ¶ ∗ [50] ‡ ‡ ‡ E-mobility E-mobility 5 [13] [29, 38] ∗ ∗ ∗
22 GW [50] ∗ ;[13] ∗ ;[13] [7] Table 3: Input parameters for modelling the load shifting process, the future scenario, investment costs, fixed costs and variable costs; ∗ own assumption; † averagedfrom [7] and [14]; ¶ approx. average from industrial processes; ¶¶ adopted from residential heat pumps; § adopted from decentralised residential heating; §§ adoptedfrom power-to-methane; ‡ adopted from air separation. Heat pump size P el [ kW ] c inv [ e / MW ]single-storey heating 2.7 115000central heating small 3.7 84000central heating medium 5.5 57000central heating large 14.2 22000 Table 4: Residential heat pumps investment costs for flexibilisation. the Grid Development Plan for 2030. We therefore used futureprojections from [14] and [7] for the considered technologies.In these studies, the industrial production, specific demands,commercial electricity demand structure and residential appli-ances energy consumption were extrapolated until 2030, basedon statistical data.For the model implementation of the future scenario, theregionalised maximum capacities of the technologies in 2018were multiplied with the relative change until 2030, x . Forthe e-mobility and PtG technologies there was only a very lowinstalled capacity in 2018. We therefore did not apply the rela-tive change factor, x , but directly used the predicted capacityto be installed in 2030, P . The numerical values are givenin Table 3.Concerning the residential, CTS and e-mobility technolo-gies, we adopted the load regionalisation for the year 2018 alsofor the 2030 scenario. The regional distribution of such loadsmainly depends on the population density, which we assumednot to significantly change until 2030. For the industrial loadshifting technologies, we also assumed a constant regional dis-tribution until 2030. Industrial sites at specific locations mightshut down and new facilities might be opened in other regionsof Germany. However, there were no reliable data available forpredicting such relocation processes and therefore we neglectedit. Regarding centralised PtH in district heating grids, as well asPtG plants, we used a di ff erent regionalisation methodology forthe future scenario as for the status quo. As described in Sec-tion 2.2.4, we used a plant specific regionalisation for the 2018status quo, based on existing resistive heaters in district heat-ing networks [36]. For the 2030 scenario, we distributed the centralised PtH plants according to the overall heat demand ofthe district heating networks in the administrative districts [23].Also for the PtG plants the load regionalisation for the statusquo is based on existing individual plants, which are researchpilot projects in most cases [40]. To model the regional distri-bution in 2030, we used the industrial gas demand per adminis-trative district [55], because according to [50] PtG will mainlybe used in the industrial sector. To model the influence of the ambient air temperature on thedemand in the PtH sector, the weather year of 2018 was alsoused for the 2030 scenario.
3. Results and Discussion
In this section we describe the results of the load shifting po-tential assessment. First, an overview of the spatial distributionof the potential is given and second, the temporal availability isassessed. Subsequently, regional cost-potential curves for theload shifting potentials are presented. Finally, the results of thispaper are compared to the literature.
The spatial distribution of the load shifting potential withinthe German administrative districts is analysed in the followingparagraphs from a geographic perspective, as well as by lookingat the frequency distribution. In Section 3.1.3 we compare theload shifting potential results with redispatch and curtailmentkey figures.
Figure 4a shows the geographical distribution of the cumu-lated maximum values for the load increase potential P max ofall considered technologies in the German administrative dis-tricts for the 2030 scenario. For better comparison each P max For more information on the regionalisation of PtH in district heating andthe PtG technologies in the future scenario refer to the supplementary material. Refer to Section 3.1.2 for a comparison of the 2018 and 2030 scenario. M W / k m (a) Cumulated load increase potential P max for all technologies. M W / k m (b) Load increase potential P max for the industrial sector.Figure 4: Geographical distribution of the load increase potential in the German administrative districts for the 2030 scenario. value was divided by the area of the administrative district. Theresults are given in MW / km , as 1000 km correspondsapproximately to the average area of the administrative districtsin Germany of 891 km .The resulting spatial distribution reflects the population den-sity with the metropolitan areas having the highest potential val-ues. These are Berlin in the east of Germany, Hamburg in thenorth, the Ruhr area in the west and Rhein-Main area in thesouth-west. Furthermore, in Germany there are about 100 moremajor cities which form an independent administrative district.Most of these districts have a population of more than 100000inhabitants, but a small territory [56], compared to the predom-inantly rural administrative districts. The resulting high areaspecific load shifting potential in the major city districts can benoted in Figure 4a by the small dark areas spread over the Ger-man territory.Figure 4b shows the load increase potential distribution foronly the industrial sector. Here, the metropolitan regions andother major cities do not dominate the potential as significantlyas in the case of the cumulated potential of all sectors. Someof the predominantly rural districts show a relatively high in-dustrial load increase potential. This can be noted by the ad-ministrative districts with a rather large territory having darkercolours compared to Figure 4a. An explanation for this findingis that industrial sites, such as cement mills or paper plants areoften located in rural areas [57, 58].For the residential, CTS, PtH and e-mobility demand sectors,no individual maps are shown here because the geographicaldistribution strongly correlates to the population density andthus corresponds to Figure 4a. For the status quo, the spatialdistribution of the power-to-gas load increase potential is set bythe locations of the power-to-gas plants in pilot projects [40]. The frequency distributions of the power and energy poten-tial values for all German administrative districts are illustratedby violin plots in Figure 5a and 5b. The distributions for loadincrease ( P max and E max ) are plotted on the positive y-axis, thedistributions for load decrease ( P min and E min ) on the negativey-axis. The potential values of the individual technologies areaggregated sector wise, as described in Section 2. The blackdashed lines denote the quartiles of the distributions, where thesecond quartile corresponds to the median.The distributions for all technologies show significant uppertails, which are caused by the largest major cities in Germanyas, e.g., Berlin or Hamburg. The distribution shapes for the res-idential appliances, the CTS sector and the PtH technologieslook relatively similar, as they are mainly driven by the num-ber of inhabitants per district. In comparison, the distributionfor the load shifting technologies in the industry sector shows asmaller upper tail, since industrial facilities are also often basedin rural areas. The distribution for PtG shows the most signif-icant upper tail, because there are only existing PtG plants infew administrative districts.For the year 2018 the PtH technologies provide the highestload increase potential, P max , with a median value of 7 . MW per administrative district, followed by residential applianceswith 5 MW and the CTS sector with 1 . MW . While the poten-tial is decreasing for residential appliances until 2030, due toimproved energy e ffi ciency, and stays approximately constantfor the industrial sector, the potential for all other energy sec-tors is increasing. The strongest increase can be observed fore-mobility with a P max median value of 0 . MW in 2018 and4 . MW in 2030. The median of the summed load increase po-tential for all sectors increases from 16 MW per administrativedistrict in 2018 to 25 MW in 2030, the maximum P max value ofall administrative districts increases from 220 MW to 390 MW .10 esidential CTS industry PtH PtG e-mobility all sectorssector10 P m i n / P m a x [ M W ] (a) Distribution of the potential values for load increase, P max (positive val-ues), and load decrease, P min (negative values). residential CTS industry PtH PtG e-mobility all sectorssector10 E m i n / E m a x [ M W h ] (b) Distribution of the potential values for energy preponing, E max (positivevalues), and energy postponing, E min (negative values).Figure 5: Distribution of the annual average load and energy shifting potential values over all German administrative districts in the di ff erent demand sectors for thestatus quo and 2030 scenario. The load decrease potential, P min , is higher than the loadincrease potential, P max , for the CTS, industrial and PtG sec-tor due to high utilisation rates. For the other sectors P min islower than P max due to low utilisation rates. The median ofthe summed load decrease potential for all sectors increasesfrom − MW per administrative district in 2018 to − MW in 2030, the maximum P min value of all administrative districtsincreases from − MW to − MW .The shapes of the distributions in Figure 5b for the energypreponing and postponing potentials, E max and E min , are simi-lar as for the power bu ff ers. Due to the higher time frame ofmanagement parameters, the potential values of the industrial,PtH, PtG and e-mobility sector are increased in comparison tothe residential and CTS sector. The median of the summedload preponing potential, E max , for all sectors increases from50 MWh per administrative district in 2018 to 130
MWh in2030. The maximum E max value of all administrative districtsincreases from 900 MWh to 2100
MWh . The magnitudes of the E min values are equal to those of the E max values, as the same ∆ t parameter is applied for both load preponing and postponing. As load shifting may be applied for avoiding grid congestionand curtailment of renewable energy sources [6], we further as-sess our results with reference to these grid management mea-sures. In the case of imminent grid congestion, one option is toapply redispatch. During this measure, the feed-in of a powerplant on one side of the potentially overloaded grid element isdecreased, while the feed-in of a power plant on the other sideis increased [59].Under current regulations conventional power plants with acapacity larger than 10 MW take part in the redispatch processin Germany [60]. Thus the calculated median value of the loadincrease potential per administrative district for all sectors of 16 MW in 2018 and 25 MW in 2030 exceeds the minimumpower limit of 10 MW . Only few administrative districts reachthis threshold value using only a single load, for example thedistrict of Heilbronn with a 100 MW resistive heater in a dis-trict heating grid [23]. Another option would be to controlsmaller shiftable loads in an aggregated manner [61], e.g., 2000medium-scale heat pumps with an installed electric capacity of5 kW . Due to changes in the legal framework , from 2021 on,power plants or storage units with a capacity larger than 100 kW will also be able to take part in the redispatch process. Thisthreshold is exceeded by the load increase potential of all ad-ministrative districts in Germany, also when regarding singleenergy sectors. Only 20 medium-scale heat pumps would needto be aggregated to reach the 100 kW threshold.German transmission grid operators are obliged to publisheach individual redispatch measure online. We averaged thepower and energy of all individual redispatch measures between2013 and 2020, yielding an average power of 230 MW and anaverage energy of 1700 MWh . Approximately 9 administra-tive districts having a load increase potential of 25 MW (me-dian value for the 2030 scenario) would need to be aggregatedto provide a power of 230 MW . To reach the average redis-patched energy of 1700 MWh , the potential of 13 administra-tive districts with a potential of 130
MWh would need to beaggregated. Consequently the average load shifting potentialof single administrative districts seems to be to small to fullycover average redispatch measures.However, load shifting may play an auxiliary role for avoid-ing redispatch of power plants and curtailment of renewableenergy sources. As an example, we consider classical 3 MW Redispatch 2.0: Amendment of the Grid Expansion Acceleration Act(NABEG), 13th of May, 2019, https: // https: // / EnWG / Redispatch MW would su ffi ce to avoidthe curtailment of 8 wind turbines. The median energy shiftingpotential of 130 MWh per administrative district would su ffi ceto avoid the curtailment of 8 wind turbines for 5 hours. We analyse the temporal distribution of the load shifting po-tential using the example of the city of Berlin for the 2030 sce-nario. The top row of Figure 6 shows the time series of thescheduled load, L , the middle row shows the load increase anddecrease potentials, P min and P max , and the bottom row showsthe energy preponing and postponing potentials, E min and E max .The left column illustrates the intraday profiles for an exem-plary cold winter working day (weather data of 1 st of March,2018). While the scheduled load of most of the considered tech-nologies is relatively constant, the residential PtH technologiesand e-mobility show significant load peaks in the morning aswell as evening hours. This leads to high load decrease po-tential during these time periods, e.g., a P min value of − MW at 06:00, and a low load increase potential, e.g., a P max value of300 MW at 06:00. Conversely, the low scheduled load duringnight time leads to a high load increase and a low load decreasepotential.As defined in Eq. 4 and 5, for the calculation of E max and E min , the scheduled load is integrated over time, using the timeframe of management, ∆ t as integration limit. This reduces theinfluence of short term load peaks and results in smoother E max and E min profiles in comparison to the profiles of P max and P min .However, there is a slight E max peak of 3500 MWh at 05:30in the morning and an E min peak of − MWh at 17:30 inthe afternoon. This can be explained by ∆ t being ≤ h formost technologies and the scheduled load being higher duringday time than during night time. Thus, most energy can bepreponed from day to the morning and postponed from day tothe evening.The right column of Figure 6 shows the load shifting poten-tial distribution during the course of the year. Due to the chang-ing space heating demand, PtH technologies are more utilisedin winter and less utilised in summer. Considering the summedpotential for all technologies, this leads to an average load de-crease value, P min , that is approx. 25% higher on a cold winterday than on a summer day. In contrast, the load increase poten-tial P max is approx. 25% higher on a summer day than on a coldwinter day.The values of E max and E min for the PtH technologies are bothincreased during winter, because the higher utilisation leads toa higher amount of energy that can be preponed and postponed.The summed potential for all technologies is twice as high on acold winter day than on a summer day. This significant di ff er-ence between summer and winter is due to the relatively highshare of PtH technologies in the E max / min potential and is caused For a discussion of the peaks in the PtH load profiles refer to Section 3.4. by the large ∆ t value for PtH, compared to the other technolo-gies. As described in Section 2.7, for each considered admin-istrative district, cost-potential curves for load shifting werecalculated. Figure 7 shows the investment cost, annualfixed cost and variable costs over the cumulated load in-crease potential P max . In order to compare regions with dif-ferent population density, we present the results for Berlin(4000 residents / km ) and the neighbouring rural districtMaerkisch-Oderland (90 residents / km ) for the 2030 scenarioas an example.In general, industrial processes and PtH in district heatinghave the lowest specific investment costs due to the high in-stalled electric capacity per facility, whereas residential appli-ances have low installed capacities and high specific investmentcosts (see also Table 3 for the numerical values). On the otherhand, industrial processes have high variable costs due to im-pairment of production processes [14], while ventilation andheating appliances have low variable cost, because of less userinterference.While the resulting overall load shifting potential for Berlinis 475 MW , it is only 29 MW in Maerkisch-Oderland, due tothe lower population density. In general, the distribution of thepotential for the di ff erent technologies is similar for Berlin andMaerkisch-Oderland. However, the most striking di ff erence be-tween the two regions is the share of the industrial processes’potential. As there is a large scale cement production facility inthe Maerkisch-Oderland district [57], but the overall load shift-ing potential is small, the industrial processes’ load shifting po-tential share is 11%. In Berlin, the share of industrial processesin the overall load shifting potential is only 1%.In order to assess the costs for load shifting, we usethe costs for the curtailment of renewable energy sourcesas a reference. According to the German Federal NetworkAgency [63], 6482 GWh of electricity from renewable energysources were curtailed in 2019, leading to compensation pay-ments of 709 . million e . Dividing the compensation paymentsby the curtailed energy yields average costs of 110 e / MWh .We further assume that load shifting for avoiding curtailmentof renewable energy sources is only used if the variable costsare lower than the average curtailment compensation costs. Allconsidered load shifting technologies meet this condition, ex-cept air separation, cement milling, wood pulping, paper pro-duction and PtG. In this case, the load shifting potential for thecity of Berlin would be reduced by 7%, which is mainly causedby the PtG technology. For the Maerkisch-Oderland district,19% of the load shifting potential would be excluded due to theshare of the cement and PtG plants.To further analyse regional influences on the load shiftingcost-potential curves, we examined the heat pump technologymore closely. As described in Section 2.7, we introduced foursize classes of heat pumps and assigned specific investmentcosts for flexibilisation for each class. Figure 8 shows the re-sulting specific investment costs over the cumulated load in-crease potential P max for Berlin and the Maerkisch-Oderland12 igure 6: Temporal distribution of the scheduled load (top row) for the city of Berlin, load increase and decrease potential (middle row) and energy preponing andpostponing potential (bottom row); Left column: cold winter day (1 st of March, 2018); Right column: Yearly time series. district. While in Berlin there is a high number of multi-familyhouses, rural areas, such as Maerkisch-Oderland are dominatedby single-family houses [23]. This leads to a 35% share of largecentral heatings in Berlin with low specific investment costs forheat pump flexibilisation of 22000 e / MW . On the contrary, inMaerkisch-Oderland there is only a 2 .
5% share of large-scaleheat pumps and 44% of heat pumps in medium central heatings,with higher specific investment costs of 57000 e / MW [23].Next, we calculated weighted cost averages, using the P max ca-pacities for each heat pump size class as weight. This results inaverage cost of 63 e / MW for Berlin and 23% higher costs of76 e / MW for Maerkisch-Oderland. To validate the results of this paper, we compared the calcu-lated values for the load increase potential, P max , with literaturereference values [7, 13, 14], as depicted in Figure 9. For sev-eral considered technologies the resulting values of this paperare higher than those in the literature and for other technologiesthe values are lower, thus there is no general under- or overes- timation. Concerning the residential appliances, the results ofthis study are 300 MW higher for washing and drying machinesand 1000 MW higher for fridges and freezers than in [14]. Thiscan be explained by the di ff erent values assumed for s util and s inc . For washing machines s util and s inc have very low values( < P max is approx. proportional to s inc − s util . There-fore di ff erences of only several percent in the input parameterslead to high deviations in the results for P max .For the industrial processes the di ff erences are between 5 − MW for all considered technologies and are probably dueto deviations in the estimated installed capacity of the pro-cesses. Regarding the PtH technologies, the results of this pa-per and [14] di ff er by 200 − MW . The largest di ff erence ispresent for resistive DHW heating. A reason for the deviation isthat in this paper it was assumed that the 25% share of resistiveDHW heaters which is equipped with a thermal storage [12] isalso suitable for load shifting, while in [14] only a 12% flexibleshare is assumed. Furthermore, we applied a utilisation rate of3% [7] and an hourly profile [23], while Steurer [14] di ff erenti-ated between two phases of the day with a constant load profile13
100 200 300 400Cumulated P max [MW] 10 I n v . c o s t s [ / M W ] Berlin 0 100 200 300 400Cumulated P max [MW] 10 A n . f i x c o s t s [ / M W ] Berlin 0 100 200 300 400Cumulated P max [MW] 10 V a r . c o s t s [ / M W h ] Berlin0 10 20 30Cumulated P max [MW] 10 I n v . c o s t s [ / M W ] Maerkisch-Oderland 0 10 20 30Cumulated P max [MW] 10 A n . f i x c o s t s [ / M W ] Maerkisch-Oderland 0 10 20 30Cumulated P max [MW] 10 V a r . c o s t s [ / M W h ] Maerkisch-Oderlandcts. cool. vent. ac.ind. ventilationpth. heat pump (cts)e-mobility pth. district heat.pth. resist. dhw. (res)pth. resist. sh. (res)pth. heat pump (res) ind. coolingres. wash. dry.res. cool. freez.ind. recycled paper pth. proc. heat (ind)ind. air separationpower-to-methanepower-to-hydrogen ind. paperind. cementind. pulp
Figure 7: Load shifting cost-potential curves for the city of Berlin (top row) and the rural district Maerkisch-Oderland (bottom row) for the 2030 scenario; Leftcolumn: Investment costs; Middle collumn: Fixed costs; Right column: Variable costs. P max [MW] 10 I n v . c o s t s [ / M W ] Berlin0 1 2 3 4 5Cumulated P max [MW] 10 I n v . c o s t s [ / M W ] Maerkisch-Oderlandhp. central heating large [14.2 kW]hp. central heating medium [5.5 kW] hp. central heating small [3.7 kW]hp. single-storey heating [2.7 kW]
Figure 8: Investment costs for the flexibilisation of residential heat pumps ofdi ff erent size classes. each and di ff ering utilisation rates.For power-to-hydrogen and e-mobility no values are given in [7] and [14] and we therefore used reference valuesfrom Pellinger et al. [13]. In [13] only the installed capacityof power-to-hydrogen plants is given, no value for the load in-crease potential. We therefore assumed a constant load profileand a utilisation rate of 44%, which yields a load increase po-tential of approx. 800 MW and is equal to the result of thisstudy. The calculated load increase potential for e-mobility isabout two times higher in this paper compared to [13]. Thereason for this is that we assumed an electric vehicle fleet of6 million cars for 2030 [50] and Pellinger et al. [13] estimated3 million electric vehicles to be existing. No literature referencevalues were found for load increase of PtH facilities coveringindustrial process heat or district heating demand, as well as forheat pumps in the CTS sector. Since the results for P min , E max and E min mainly depend on the same input parameters as P max ,we did not carry out a separate validation for these potentials.Regarding the temporal distribution of the load shifting po-tential, there are significant peaks in the load increase and de-crease time series for the residential space heating and DHWpower-to-heat technologies. The reason for this is that theutilised daily load profiles [23] are based on measurements of14 Load increase potential P max [ MW ]res. wash. dry.res. cool. freez.cts. cool. vent. ac.ind. air separationind. cementind. pulpind. paperind. recycled paperind. coolingind. ventilationpth. proc. heat (ind)pth. heat pump (res)pth. resist. sh. (res)pth. resist. dhw. (res)pth. district heat.pth. heat pump (cts)power-to-methanepower-to-hydrogene-mobility this studyreference Figure 9: Validation of the results for the annual average load increase potentialvalues, P max , in all of Germany for the 2030 scenario with literature referencevalues; The following validation references were used for the considered tech-nologies: res. wash. dry. [14], res. cool. freez. [14], cts. cool. vent. ac. [7],ind. air separation [14], ind. cement [14], ind. pulp [14], ind. paper [14], ind.recycled paper [14], ind. cooling [14], ind. ventilation [7], pth. heat pump(res) [14], pth. resist. sh. (res) [14], pth. resist. dhw. (res) [14], power-to-hydrogen [13], e-mobility [13]. the heat flow and return at the heat exchangers and representthe actual demand of the residents. Since the considered elec-tric heating devices are equipped with thermal storage [64], theelectricity consumption can be decoupled from the actual de-mand. Thus, electric heating devices often run at rather constantload and increased demand during night time, to support the op-eration of conventional power plants, which is incentivised bytari ff design [65]. Such a load profile might be used as referencescheduled load for applying additional load shifting, e.g. foravoiding curtailment of renewable energy sources. This wouldresult in smoother profiles for the load increase and decreasepotentials. However, we did not use such a load profile as areference, as it already includes a preceding load shifting, com-pared to the actual demand of the residents.Furthermore, the utilised input parameters given in Table 3for modelling demand response and the associated costs arefraught with uncertainty. As load shifting is not widely usedcurrently, the parameters need to be determined by small-scalepilot projects and surveys [14]. E.g. for heat pumps, Steurer[14] estimates the uncertainty for the time frame of manage-ment parameter, ∆ t , at ±
4. Conclusion and Outlook
One option to avoid the curtailment of renewable energies isto cover excess feed-in by load shifting. As curtailment is oftena local phenomenon, in this work we determined the regionalload shifting potential and the associated costs for the 401 Ger-man administrative districts, considering 19 suitable technolo-gies. The results are provided with a temporal resolution of 15 minutes and a status quo analysis for 2018, as well as a 2030scenario are considered.In contrast to other studies in this field, all data and the de-veloped source code are published open source. We highlightedthe load shifting potential of sector coupling technologies bytaking into account power-to-heat, power-to-gas and e-mobilityin addition to the considered conventional loads. Further, weput a special focus on the regionalisation of the residentialbuilding stock, considering more than 700 building types.The highest load shifting potential is provided by the power-to-heat technologies, the lowest by the considered industrialprocesses suitable for load shifting. The strongest growth ofthe potential from 2018 to 2030 can be observed for power-to-gas and e-mobility. The spatial potential distribution is mostlygoverned by the population density. For industrial processes,there are also relatively high potential values in rural areas.For the 2030 scenario, the load increase potential values rangefrom 5 − MW per administrative district and the medianvalue is 25 MW . In comparison, the lower power threshold forpower plants to take part in redispatch is 10 MW and the aver-age power of redispatch measures in the German transmissiongrid is 230 MW . Load shifting is thus not a real alternativeto redispatch of power plants, but may play an auxiliary role.When compared to curtailment, the median load increase po-tential value per administrative district would su ffi ce to avoidthe curtailment of 8 wind turbines with 3 MW installed capac-ity.The temporal distribution of the load increase and decreasepotentials shows significant peaks in the morning and afternoonhours, caused by the demand time series of PtH and e-mobility.Due to the changing space heating demand in the course of theyear, load decrease potentials are approx. 25% higher on a coldwinter day than on a summer day.Industrial processes, power-to-gas and power-to-heat in dis-trict heating have the lowest load shifting investment cost, dueto the largest installed capacities per facility. Ventilation, cool-ing and heating appliances have the lowest variable costs dueto the least user interference. Further distinguishing betweendi ff erent size classes of the installed capacity of heat pumpsleads to 23% lower average investment costs for heat pumpflexibilisation in the city of Berlin compared to the rural dis-trict of Maerkisch-Oderland. The variable costs of most con-sidered load shifting technologies remain under the averagecompensation costs for curtailment of renewable energies of110 e / MWh .As the load shifting potentials time series determined in thisstudy are published open source, they can be used by otherresearchers as boundary conditions for load shifting dispatchin energy system models. The provided investment, fixed andvariable costs for load shifting can be further used for an eco-nomic assessment considering the full life time of the technolo-gies. For industrial process heat, power-to-gas and power-to-heat in district heating, detailed technical studies are requiredfor deriving more reliable load shifting parameter values.15 cknowledgements
The first author gratefully acknowledges the financial supportprovided by the Foundation of German Business (sdw) througha PhD scholarship. The work of the second and third authorwas carried out as part of the enera project, which is funded bythe Federal Ministry of Economic A ff airs and Energy (BMWi,grant no. 03SIN317). Further, the authors thank Hans ChristianGils, Karl-Kien Cao and Jan Jebens for the fruitful discussionson regionalisation and load shifting, as well as Niklas Wul ff forvaluable hints regarding e-mobility modelling. Appendix A. Supplementary Material
Supplementary data associated with this article can be foundat: https: // doi.org / / zenodo.3988921. References [1] R. E. Sims, Renewable energy: a response to climate change, Solarenergy 76 (2004) 9–17.[2] REN21, Renewables 2019 global status report, Online. , 2019.[3] T. Luz, P. Moura, 100 percent renewable energy planning with comple-mentarity and flexibility based on a multi-objective assessment, AppliedEnergy 255 (2019) 113819.[4] T. Mueller, D. Moest, Demand response potential: Available whenneeded?, Energy Policy 115 (2018) 181–198.[5] J. Valdes, A. B. P. Gonz´alez, L. R. Camargo, M. V. Fen´andez, Y. M.Macia, W. Dorner, Industry, flexibility, and demand response: applyingGerman energy transition lessons in Chile, Energy Research & SocialScience 54 (2019) 12–25.[6] P. S. Moura, A. T. De Almeida, The role of demand-side management inthe grid integration of wind power, Applied Energy 87 (2010) 2581–2588.[7] H. C. Gils, Balancing of intermittent renewable power generation by de-mand response and thermal energy storage, Ph.D. thesis, University ofStuttgart, 2015.[8] M. Paulus, F. Borggrefe, The potential of demand-side management inenergy-intensive industries for electricity markets in Germany, AppliedEnergy 88 (2011) 432–441.[9] I. Stadler, Demand Response- Nichtelektrische Speicher f¨ur Elek-trizit¨atsversorgungssysteme mit hohem Anteil erneuerbarer Energien,Ph.D. thesis, Universit¨at Kassel, 2005.[10] M. Klobasa, Dynamische Simulation eines Lastmanagements und Inte-gration von Windenergie in ein Elektrizit¨atsnetz auf Landesebene unterregelungstechnischen und Kostengesichtspunkten, Ph.D. thesis, Univer-sit¨at Karlsruhe, 2007.[11] C. Elberg, C. Growitsch, F. H¨o ffl er, J. Richter, A. Wambach, Unter-suchungen zu einem zukunftsf¨ahigen Strommarktdesign, Studie im Auf-trag des BMWi, durchgef¨uhrt vom Energiewirtschaftliches Institut an derUniversit¨at K¨oln (2012).[12] R. Apel, T. Aundrup, B. Buchholz, H. Domels, S. Funke, T. Gesing,W. Glaunsinger, P. Gronstedt, F. Heins, A. H¨osle, et al., Ein notwendi-ger Baustein der Energiewende: Demand Side Integration, Lastver-schiebungspotenziale in Deutschland. Kurzfassung (Verband der Elek-trotechnik Elektronik Informationstechnik eV (VDE), Hrsg.). Frankfurta. M.: Energietechnische Gesellschaft im VDE (ETG) (2012).[13] C. Pellinger, T. Schmid, A. Regett, A. Gruber, J. Conrad, K. Wachinger,S. Fischhaber, Merit Order der Energiespeicherung im Jahr 2030-Hauptbericht, M¨unchen: Forschungsstelle f¨ur Energiewirtschaft eV (FfE)(2016).[14] M. Steurer, Analyse von Demand Side Integration im Hinblick auf einee ffi ziente und umweltfreundliche Energieversorgung, Ph.D. thesis, Uni-versit¨at Stuttgart, Institut f¨ur Energiewirtschaft und Rationelle Energiean-wendung, 2017. [15] L. S¨oder, P. D. Lund, H. Koduvere, T. F. Bolkesjø, G. H. Rossebø,E. Rosenlund-Soysal, K. Skytte, J. Katz, D. Blumberga, A review ofdemand side flexibility potential in Northern Europe, Renewable and Sus-tainable Energy Reviews 91 (2018) 654–664.[16] L. von Bremen, M. Buddeke, D. Heinemann, A. Kies, D. Kleinhans,C. Krueger, F. Merten, M. Preute, S. Samadi, T. Vogt, L. Wien-holt, RESTORE 2050 – Regenerative Stromversorgung und Spe-icherbedarf im Jahr 2050, Projektabschlussbericht: Ergebnisse undHandlungsempfehlungen, Teilbericht D13 + D14, 2016. FKZ 03SF0439,gef¨ordert durch das Bundesministerium f¨ur Bildung und Forschung(BMBF).[17] German Federal Statistical O ffi ce, DESTATIS, NUTS classifi-cation, the hierarchical categorisation of EU territories and re-gions, 2020. .[18] Schucht et al., Netzentwicklungsplan Strom 2030, Version 2017 - ZweiterEntwurf der ¨Ubertragungsnetzbetreiber, 2017.[19] I. Stadler, Ein gigantisches Speicherpotenzial, Solarzeitalter 1 (2008)60–64.[20] M. Victoria, K. Zhu, T. Brown, G. B. Andresen, M. Greiner, The roleof storage technologies throughout the decarbonisation of the sector-coupled European energy system, Energy Conversion and Management201 (2019) 111977.[21] W. Medjroubi, U. P. M¨uller, M. Scharf, C. Matke, D. Kleinhans, Opendata in power grid modelling: new approaches towards transparent gridmodels, Energy Reports 3 (2017) 14–21.[22] D. Kleinhans, Towards a systematic characterization of the potential ofdemand side management, Online. https://arxiv.org/abs/1401.4121 , 2014.[23] W. Heitkoetter, W. Medjroubi, T. Vogt, C. Agert, Regionalised heatdemand and power-to-heat capacities in Germany–An open dataset forassessing renewable energy integration, Applied Energy 259 (2020)114161. https: // doi.org / / j.apenergy.2019.114161.[24] D. Schmidt, NUTS-3 Regionalization of Industrial Load Shifting Poten-tial in Germany using a Time-Resolved Model, Master Thesis, Universityof Oldenburg, 2019. https://elib.dlr.de/134116/ .[25] Bundesministerium f¨ur Wirtschaft und Energie, Zahlen und Fakten En-ergiedaten, 2018.[26] German Federal Statistical O ffi ce, DESTATIS, GENESIS Online, Tabelle43531-0001, Stromverbrauch der Betriebe im Verarbeitenden Gewerbe,2018. https: // ffi ciency and energy saving potential in industry and on possible policymechanisms, ICF Consulting Limited, London, 2015.[28] Arbeitsgemeinschaft Energiebilanzen, , Satellitenbilanz Erneuerbare En-ergietr¨ager, 2017.[29] Nitsch, J., Pregger, T., Naegler, T., Heide, D., de Tena, D. Luca, Trieb,F., Scholz, Y., Nienhaus, K., Gerhardt, N., Sterner, M. et al., Langfrist-szenarien und Strategien f¨ur den Ausbau der erneuerbaren Energien inDeutschland bei Ber¨ucksichtigung der Entwicklung in Europa und global,Schlussbericht im Auftrag des BMU, bearbeitet von DLR (Stuttgart),Fraunhofer IWES (Kassel) und IfNE (Teltow), 2012.[30] German Federal Motor Transport Authority, Bestand an Kraftfahrzeu-gen und Kraftfahrzeuganh¨angern nach Zulassungsbezirken, 2019.https: // ffi ce, DESTATIS, Regionalatlas Deutsch-land, Indikatoren des Themenbereichs ”Verdienste und Einkommen”,Tabelle Al1601 (Verf¨ugbares Einkommen je Einwohner), 2017.[32] Oberascher, Claudia, Stromverbrauch im Haushalt, Technical Report,BDEW Bundesverband der Energie- und Wasserwirtschaft e.V., Berlin,2016.[33] German Federal Statistical O ffi ce, DESTATIS, Erwerbst¨atige nachWirtschaftszweigen - Jahresdurchschnitt 2017- regionale Tiefe: Kreiseund krfr. St¨adte, Erwerbst¨atigenrechnung des Bundes und der L¨ander,2018. .[34] German Federal Statistical O ffi ce, DESTATIS, Betriebe und Besch¨aftigtenach Wirtschaftszweigen 2017, Tabelle 42111-02-03-4-B, 2018. .[35] German Federal Statistical O ffi ce, DESTATIS, Stromverbrauch derBetriebe im verarbeitenden Gewerbe, 2018, Tabelle 43531-0001, 2019. .[36] A. Christidis, E. Mollenhauer, G. Tsatsaronis, G. Schuchardt, S. Holler,D. B¨ottger, T. Bruckner, EnE ff -W¨arme: Einsatz von W¨armespeichern undPower-to-Heat-Anlagen in der Fernw¨armeerzeugung, AGFW, Frankfurtam Main, Germany (2017).[37] German Federal Statistical O ffi ce, DESTATIS, Bruttoinlandsprodukt,Bruttowertsch¨opfung in den kreisfreien St¨adten und Landkreisen derBundesrepublik Deutschland, 1992 und 1994 bis 2016, Reihe 2,Kreisergebnisse Band 1, 2017. .[38] D. Luca de Tena, Large scale renewable power integration with electricvehicles: long term analysis for Germany with a renewable based powersupply, Ph.D. thesis, University of Stuttgart, 2015.[39] infas, DIW, Mobilit¨at in Deutschland, Bonn / Berlin, 2003.[40] M. Thema, F. Bauer, M. Sterner, Power-to-gas: Electrolysis and metha-nation status review, Renewable and Sustainable Energy Reviews 112(2019) 775–787.[41] C. F¨unfgeld, R. Tiedemann, Anwendung der repr¨asentativen VDEW-Lastprofile: step-by-step, VDEW, 2000.[42] Lange, M., Zobel, M., Schlussbericht zum Vorhaben: Erstellungneuer Referenzlastprofile zur Auslegung, Dimensionierung undWirtschaftlichkeitsberechnung von Hausenergieversorgungssys-temen (NOVAREF), DLR-Institut f¨ur Vernetzte Energiesysteme,F¨orderkennzeichen BMWi 03FS14013, 2017.[43] G. BDEW, VKU, Abwicklung von Standardlastprofilen Gas, 2011.[44] German Meteorological Service, CDC - Climate Data Centre, 2018.https: // cdc.dwd.de / portal / .[45] M. Hellwig, Entwicklung und Anwendung parametrisierter Standard-Lastprofile, Ph.D. thesis, Technical University of Munich, 2003.[46] O. Ruhnau, L. Hirth, A. Praktiknjo, Time series of heat demand and heatpump e ffi ciency for energy system modeling, Scientific data 6 (2019)1–10.[47] I. Sta ff ell, D. Brett, N. Brandon, A. Hawkes, A review of domestic heatpumps, Energy & Environmental Science 5 (2012) 9291–9306.[48] Bundesverband W¨armepumpe, W¨armepumpen Bestand in Deutschland,2016.[49] Bundesverband W¨armepumpe, BWP-Branchenstudie 2018, Marktanal-yse – Szenarien – Handlungsempfehlungen, 2018.[50] Bundesnetzagentur, Genehmigung des Szenariorahmens Netzentwick-lungsplan Strom 2030, 2019.[51] U. Albrecht, U. B¨unger, J. Michalski, W. Weindorf, J. Zerhusen,F. Borggrefe, H. C. Gils, T. Pregger, F. Kleiner, J. Pagenkopf, et al.,Kommerzialisierung der Wassersto ff technologie in Baden-W¨urttemberg-Rahmenbedingungen und Perspektiven (2016).[52] S. J. Darby, E. McKenna, Social implications of residential demand re-sponse in cool temperate climates, Energy Policy 49 (2012) 759–769.[53] H. C. Gils, Economic potential for future demand response in Germany–Modeling approach and case study, Applied Energy 162 (2016) 401–415.[54] A. Bloess, W.-P. Schill, A. Zerrahn, Power-to-heat for renewable energyintegration: A review of technologies, modeling approaches, and flexibil-ity potentials, Applied Energy 212 (2018) 1611–1626.[55] F. Kunz, J. Weibezahn, P. Hauser, S. Heidari, W. Schill, B. Felten,M. Kendziorski, M. Zech, J. Zepter, C. von Hirschhausen, et al., Ref-erence Data Set: Electricity, Heat, and Gas Sector Data for Model-ing the German System, 2017. https://doi.org/10.5281/zenodo.1044462 .[56] Statistische ¨Amter des Bundes und der L¨ander, Zenus 2011, 2013.https: // , 2019.[58] VDP, Standorte und Produktionsst¨atten der VDP- Mitgliedsunternehmen,Online. , 2017.[59] A. F. K. Kamga, Regelzonen¨ubergreifendes Netzengpassmanagement mitoptimalen Topologiemaßnahmen, Wuppertal, 2009.[60] L. Hirth, I. Schlecht, B. Maurer, Christoph und Tersteegen, Kosten- oderMarktbasiert? Zuk¨unftige Redispatch-Bescha ff ung in Deutschland, 2019.[61] K. Paridari, A. Parisio, H. Sandberg, K. H. Johansson, Demand responsefor aggregated residential consumers with energy storage sharing, in:2015 54th IEEE conference on decision and control (CDC), IEEE, pp.2024–2030.[62] L. Hirth, S. M¨uller, System-friendly wind power: How advanced windturbine design can increase the economic value of electricity generatedthrough wind power, Energy Economics 56 (2016) 51–63.[63] Bundesnetzagentur, Quartalsbericht Netz- und Systemsicherheit -Gesamtes Jahr 2019, 2020.[64] G. P. Henze, C. Felsmann, G. Knabe, Evaluation of optimal control foractive and passive building thermal storage, International Journal of Ther-mal Sciences 43 (2004) 173–183.[65] S. J. Darby, Smart electric storage heating and potential for residentialdemand response, Energy E ffi ciency 11 (2018) 67–77.ciency 11 (2018) 67–77.