Sharing Mobile and Stationary Energy Storage Resources in Transactive Energy Communities
aa r X i v : . [ ee ss . S Y ] O c t Sharing Mobile and Stationary Energy StorageResources in Transactive Energy Communities
Pedro Moura
ISR, Electrical and Computer Eng.University of Coimbra
Coimbra, [email protected]
Uday Sriram
Electrical and Computer Eng.Carnegie Mellon University
Pittsburgh, [email protected]
Javad Mohammadi
Electrical and Computer Eng.Carnegie Mellon University
Pittsburgh, [email protected]
Abstract —Most power systems are increasingly based ondistributed energy resources, leading to a strong impact onthe electrical grid management. In the case of Portugal, thecontribution of renewable energy sources to the electricitygeneration portfolio is already high and the objective is toachieve 100% by 2050. Most of the new renewable generationcapacity will be ensured by distributed photovoltaic generationinstalled in buildings and therefore the inherent intermittenceof photovoltaic output combined with a mismatch with de-mand profile will challenge the operation and resiliency of theelectrical grid. Addressing these issues requires managementat the community level and the leveraging of spatio-temporalflexibility of controllable energy resources, such as energy storageresources. This is recognized by the regulators in Portugal and therecent renewable generation self-consumption legislation enablesgeneration-surplus trading in communities. Implementing intra-community trading and utilizing the potentials of renewablegeneration requires oversight and coordination at the communitylevel in the context of transactive energy systems. This paperfocuses on addressing energy sharing through a transactiveenergy market in community microgrids, using stationary andmobile energy storage as flexibility resources. The proposedframework considers public and commercial buildings with on-site battery storage and numerous electric vehicle chargingstations. The formulation is assessed using real data from acommunity of buildings on a Portuguese University campus. Theresults showcase that the proposed method achieves an increasein renewable self-consumption at building and community levels,as well as a reduction in electricity costs.
Index Terms —Community Microgrid, Transactive EnergyMarket, Distributed Energy Resources, Battery Storage, ElectricVehicles. N OMENCLATURE
Inputs ∆ h Time step ( hour ) C P Baseline parking tariff for EVs parking ( e /h ) C C ( h ) , Tariff for the charging/discharging of EVs at C D ( h ) time step h ( e /h ) C F Reward for EV charging flexibility ( e /h ) C EG ( h ) , Tariff for power exported/imported to/from C IG ( h ) the grid at time step h ( e /kW h ) C G ( h ) Tariff for grid use between buildings ( e /kW h ) Support for this research was provided by the Fundac¸ ˜ao para a Ciˆencia ea Tecnologia (Portuguese Foundation for Science and Technology) throughthe Carnegie Mellon Portugal Program and by OE - national funds ofFCT/MCTES (PIDDAC) under project UID/EEA/00048/2019. t bP,n Total parking period of EV n in building b ( h ) t b + R,n , Charging/discharging period requested by EV t b − R,n owner n in building b ( hour ) L b + ( h ) , Positive/negative net electricity load in building L b + ( h ) b at time step h ( kW ) P b + A. Motivation The decarbonization and expansion of distributed energyresources are clear drivers of change in the electric powersystem, which is increasingly based on distributed, intermit-tent, and non-dispatchable renewable sources. Portugal alreadyas 55% of the electricity generation ensured by renewablesand aims at achieving 100% renewable electricity generationby 2050 [1]. This will impact the future of an integratedgrid at all scales, but mainly in buildings and communities,since 25% of the capacity will be ensured by decentralizedphotovoltaic (PV) generation. However, in most buildings,there is a high mismatch between the local PV generationand demand profiles, leading to the need to export to the grida significant part of the locally generated energy, even thoughthe same amount is later imported back for local consumption[2]. This creates challenges for the electrical grid managementand leads to economic losses to the end-user [3].In this context, it will be fundamental to have a resilienttransactive grid, being the integration and management ofnew technologies to provide flexibility crucial to achieve suchan objective. At building and community levels, distributedenergy storage with Battery Storage (BS) systems has emergedas an attractive solution for this new paradigm due to itsdecreasing costs and increasing efficiency and reliability.Simultaneously, the transport sector with Electric Vehicles(EV) is increasingly an important consumer of electricity andPortugal aims at achieving 70% electrification of transportsby 2050 [1]. Therefore, EVs parked in buildings can alsobe used as flexible resources in transactive energy systems,adjusting the charging period with the Building-to-Vehicle(B2V) system, or used as energy storage resources by injectinginto the building part of the stored energy with the Vehicle-to-Building (V2B) system [4]. Additionally, the new legislationfor the self-consumption of renewable generation in Portugalenables the establishment of renewable energy communities,in order to share and trade the renewable generation surplus.Therefore, an aggregated management at the community levelwill be required to optimize the matching between renewablegeneration and demand in a transactive energy context. B. Related Works There is a vast body of works proposing methodologies toimplement the participation of buildings in transactive energymarkets, and the management of flexibility resources.The implementation of transactive mechanisms for the man-agement of EVs and energy storage is mainly considered inresidential buildings. In [5] a transactive energy control forresidential prosumers with BS and EVs is proposed. In [6] thecase of EVs participating in a retail double auction electricityregulation market is considered, and in [7] a two-stage optimalcharging scheme based on transactive control is proposed.However, in residential buildings, the all flexibility resources(including the EVs) belong to the building and there are noeconomic transactions between entities in order to use suchresources.Some works have also considered commercial buildingsin transactive energy markets. In [8] the characteristics ofcommercial buildings and end uses are explored to determinefactors supporting the feasibility of participation in transactiveenergy systems. In [9] a transactive control market structurefor commercial building HVAC systems is presented and in [10] a passive transactive control strategy was applied toestimate the peak demand reduction potential and energysavings of a building. However, such works only considerdemand flexibility, without the use of energy storage resources.Other works consider the economic relationship betweenbuildings and EV users and. In [11] an office building withPV and EVs is considered with the objective of minimizingenergy costs and in [12] a building with renewable generationand storage and EVs charging directly with the generatedenergy is considered in order to ensure the minimizing costsand greenhouse gas emissions. In [13] several commercialbuildings with EV charging stations are considered with theobjective of minimizing the charging costs, as well as theglobal energy costs in the building. However, such works con-sider a direct trade of electricity between buildings and EVswith payments associated with the electric power flow whichis not allowed for nearly all entities by the actual legislation inmost countries. In [14], a first approach based on the parkingcosts to regulate the economic relationship between buildingand EV user is proposed, but without considering aggregationat the community level.In [15] a transactive real-time EV charging managementscheme is proposed for the building energy managementsystem of commercial buildings with PV generation andEV charging. However, such an approach requires complexinformation from the EV users that is not easily obtained inreal scenarios and does not consider the optimization at thecommunity level. In [16], a community is considered, usingEVs as flexibility resources, but without considering energystorage and without implementing transactive energy marketmechanisms in the aggregation at community level. C. Contribution The main contribution of his work is the design of a trans-active energy market for community microgrids constituted bylarge public and commercial buildings, using BS and EVs asflexibility resources. Therefore, a formulation is proposed toestablish a transactive energy market at the community level,using price signals for the energy injected or consumed fromthe community, in order to give incentives for the aggregatedmatching between demand and PV generation while ensuringthe minimization of electricity costs. Such management isnot only ensured with transactions between buildings, butalso with flexibility resources in buildings, such as BS andV2B/B2V systems. Therefore, the formulation implementsthe management of such flexibility resources at the buildinglevel. Since the formulation considers the case of large publicand commercial buildings with parking lots, where EVs andbuildings do not belong to the same entity, a transactivemarket between buildings and EV users is also established.The economic relationship between EV users and buildingsis based on the parking time and added value services for thecharging in order to minimize the monitoring requirements andcomply with Portuguese legislation that does not allow directtrading of electricity between buildings and EVs. . Paper Organization The remainder of the paper is structured as follows. Sec-tion II describes the problem formulation and Section IIIpresents the data and scenarios. The simulation results arepresented and discussed in Section IV. Finally, Section Vsummarizes the paper, emphasizing its main conclusions.II. P ROBLEM F ORMULATION A. Objective Function The objective of the proposed formulation (1) is the min-imization of total costs at the community level during theassessed period (i.e., H ). Therefore, it is considered a com-munity with B buildings, all with PV generation, BS and EVsresources. The total costs in each building take into accountnot only the electricity costs, but also the income due to theparking and charging of N EVs. min B X b =1 H X h =1 C bE ( h ) − N X n =1 C bEV ( n ) ! (1) The electricity cost in each building b (2) is associatedwith the net electricity load and considers the cost/incomeof energy drawn/injected from the community or from thegrid. Therefore, such cost is influenced by the baseline netelectricity load in each building, as well as the respectivepower flows with the BS, EVs and the community. C bE ( h ) = ∆ h · h P b − c ( h ) · C IC + P b + c ( h ) · C EC + L b + ( h ) − P b − c ( h ) − P b − BS ( h ) − N X n =1 P b − EV,n ( h ) ! C IG ( h )+ L b − ( h ) − P b + c ( h ) − P b + BS ( h ) − N X n =1 P b + EV,n ( h ) ! C EG ( h ) (2) The parking and charging service of EVs considers aparking scheme that links the parking and charging periodsrequested by the EV user, as well as the allowed dischargingperiod with the electricity values, as presented in detail in [14].Therefore, tariffs associated with the parking and chargingperiods are considered, as well as tariffs to provide rewardsby the discharging and flexibility periods. The flexibility tariffensures a reward for the idle period (i.e., parking periodwithout charging or discharging), since longer idle periodsallow higher scheduling flexibility. The discharging perioddefined by the user is a maximum value and the building canuse a lower value. Additionally, the charging period can behigher than the charging period requested by the user in orderto compensate for any used discharging period. Therefore, thetotal parking costs (3), for each EV n in building b , depend onthe parking period, used periods for charging and dischargingin each time step, and the total charging and dischargingperiods over all time steps. C bEV ( n ) = t bP,n · C bP + ( t bP,n − t b + T,n − t b − T,n ) · C F + H X h =1 (cid:16) t b + U,n ( h ) · C C ( h ) (cid:17) + H X h =1 (cid:16) t b − U,n ( h ) · C D ( h ) (cid:17) (3) Equations (4) and (5) derive the net used charging anddischarging periods in each time step and calculate the totalperiods over all time steps H . t + U,n ( h ) = P + EV,n ( h ) P + The proposed formulation is subject to constraints relatedto the flexibility resources, as well as with the management ofthe community. 1) Electric Vehicles: The periods required by the EV usershould be ensured and therefore, the requested charging pe-riod, added by the compensation of the discharging periodmust be achieved until the end of the parking period (6). How-ever, the charging and discharging periods requested by the EVuser were defined based on the maximum charging/dischargingpower. Therefore, such periods include a correction propor-tional to the ratio between the average and maximum power. t b + T,n = t b + R,n P b + 2) Battery Storage: The charging and discharging power ofthe batteries is limited by the maximum (10) and minimum(11) SoC of the batteries (12), as well as by the maximumcharging or discharging power (13). Additionally, it is notpossible to simultaneously charge and discharge each battery(14). P b + BS ( h ) · ∆ h · η bBS,n ≤ (cid:0) − S bBS ( h − 1) + S b ATA AND S CENARIOS A. Buildings The data used to assess the proposed formulation is fromthe Department of Electrical and Computer Engineering atthe University of Coimbra (Portugal), a building with about10,000 m and yearly electricity consumption of about500 M W h/year . The building is equipped with a PV systemwith 79 kW p , ensuring about 16% of the actual electricitydemand [17]. The actual generation level does not lead to afrequent renewable generation surplus and therefore a futurescenario with a larger PV generation system able to ensure50% of the demand was considered, being therefore the PVgeneration data adjusted to the new PV capacity.The proposed formulation was simulated for 24 hoursperiods and in order to ensure scenarios with intermediate consumption and generation, representative days from Marchwere selected. To represent different buildings, with similarcharacteristics in one community, four different days fromdifferent weeks in March were selected to represent the fourbuildings of the community, as presented in Fig. 1. As canbe seen, such data assumes that there are simultaneouslybuildings with a PV generation surplus and others with a PVgeneration deficit. In a specific community, the variation of thePV generation presents a high correlation, but the variationsof demand may not be correlated, being possible to assumethe existence of different net electricity load profiles. h:m -150-100-50050100150 k W B1B2B3B4 Fig. 1. Net electricity load in the four considered buildings B. Tariffs The actual tariffs in the reference building were used todefine the considered tariffs for the electricity imported orexported to the grid. The electricity imported from the gridconsiders an average cost equal to the average cost in thebuilding (122.8 e /M W h ), but with a variation proportionalto the average profile of prices in the wholesale market inMarch. The electricity exported to the grid considers a flattariff with 90% of the monthly average of the wholesale market(-35.8 e /M W h ), as defined by the Portuguese regulation(Fig. 2). In the electricity trading in the community, theconsidered tariff for the grid use between buildings is a flattariff of 50 e /M W h . h:m -50050100150 ExportImport Fig. 2. Tariffs for the electricity imported and exported to the grid egarding the tariffs for the parking and charging of EVs,flat tariffs of 0.5 e /h and -0.5 e /h , respectively, wereconsidered for the parking and flexibility. The charging anddischarging of EVs considered tariffs with a variation propor-tional to the tariff for the electricity imported from the gridand an average of 2 e /h and -3 e /h , respectively. C. Electric Vehicles The EVs for the simulations consider a parking periodmainly concentrated between 8 a.m. and 8 p.m. . As presentedin Tab. I, requirements of 30 EVs were generated with anaverage of 8:00 hours , 2:00 hours and 0:45 hours forthe parking, charging and discharging periods, respectivelyand a small standard deviation in order to ensure uniformrequirements. Such values are aligned with the typical parkingrequirements in the reference building. From the 30 availableEV profiles, six EVs were randomly selected to be parkedin each building. The EV chargers used in the buildingsconsidered a maximum charging/discharging power of 10 kW and efficiency of 93%. TABLE IP ARKING REQUIREMENTS Period AVG STD MIN MAXParking 8:00 1:00 6:15 11:00Charging 2:00 0:30 1:15 3:00Discharging 0:45 0:15 0:00 1:15Start 9:30 0:45 8:00 10:15 D. Battery Storage The reference building has a BS system with lithium-ionbatteries, ensuring a total storage capacity of about 30 kW h and inverters with a charging/discharging power of 15 kW .As in the case of the PV generation, a future scenariowas also considered, being used the same adjustment factor.Therefore, a storage capacity of 90 kW h and 45 kW ofcharging/discharging power was considered.IV. S IMULATION R ESULTS The formulation was simulated for the presented data andscenarios, considering optimization at building and communitylevels. For example, Fig. 3 presents the results for building 2,with the net electricity load for the baseline scenario (withoutthe use of flexibility resources) and scenarios including theimpact of EVs and BS with individual and community man-agement of buildings. The use of the flexibility resources inthe scenarios with individual and community management ispresented in more detail in Fig. 4.As can be seen, the BS and EVs preferentially charge inperiods of negative net electricity load (PV generation surplus)and discharge in periods of positive net electricity load (PVgeneration deficit). Additionally, the BS also charges duringthe night taking advantage of the period with lower tariffsfor the energy imported from the grid, being such energy h:m -150-100-50050100150 k W BaseloadCommunityIndividual Fig. 3. Net electricity load for building 2 with individual and communitymanagement h:m -80-60-40-20020406080 k W EVComEVIndESComESInd Fig. 4. Power flow between EVs, BS and building 2 partially used to ensure the charging of the first EVs in themorning. The presented net electricity load corresponds tothe power flow with the grid, being therefore influenced bythe power flow with the community in the case of com-munity management. Therefore, in the case of communitymanagement, it was possible to compensate for all periods ofnegative net electricity load which was not possible with theindividual management during a short period. The availabilityof an additional flexibility resource (the power flow withthe community) in the community management justifies theslightly different profiles for the use of BS and EVs.In the case of community management, there are buildingsimporting and others exporting to the community, as canbe seen in Fig. 5. As a result of the market establishedbetween such buildings, the tariff for the energy exported tothe community was -68.99 e /M W h and the tariff for theenergy imported from the community was 121.48 e /M W h .The costs achieved in the simulated scenarios are presentedin Tab. II. It should be noted that, when compared with thebaseline, the individual and community scenarios require anenergy consumption increase of 29.5% due to the demandassociated with the charging of EVs and storage losses.However, due to the use of flexibility ensured by EVs and BS,such higher demand was ensured at a lower cost. Additionally,he management at the community level was able to ensurea 3% reduction in costs. The objective function takes intoaccount not only the electricity costs, but also the costs paidby EV users, and due to the profit ensured by the designedcharging scheme, the total costs relative to the baseline werereduced by 27%. h:m -150-100-50050100150 k W B1B2B3B4 Fig. 5. Power flow between each building and the communityTABLE IIC OSTS BY BUILDING AND SCENARIO ( e )Buil. Base. Individual Community C E C E C EV Obj. C E C EV Obj.1 129.6 122.9 -31.5 91.4 121.4 -31.4 89.92 140.0 141.2 -39.1 102.0 139.3 -39.1 100.23 201.9 214.5 -33.0 181.5 202.3 -32.8 169.54 90.0 82.0 -31.0 51.1 82.0 -30.9 51.1Total 561.5 560.6 -134.6 426.0 545.0 -134.3 410.7 V. C ONCLUSIONS This paper proposes a novel method to establish a realistictransactive energy market for energy sharing in Portugal.The energy trade is envisioned at the community microgridlevel and enables mostly by large public and commercialbuildings with on-site batteries and EV charging stations.The proposed method also establishes a transactive energymarket between buildings and EV users in which parking timeand added value services (stemming from charging) are thecurrency that facilitates economic relationships. The proposedmethod is aligned with the Portuguese legislation that doesnot allow direct electricity trading between buildings and EVswhile allowing for energy-surplus sharing between buildingsin renewable energy communities.The formulation was simulated for a building communitylocated at the campus of the University of Coimbra. The resultsdemonstrate that such formulation is able to take advantageof the management of energy storage resources to ensurean increased matching between the demand and local PVgeneration, as well as lower costs. The proposed method yieldsa higher impact when the management is implemented at the community level, therefore highlighting the benefits ofaggregation ensured by the transactive energy community.R EFERENCES[1] Portugal, “Roadmap for carbon neutrality 2050,” Portuguese Republic,Tech. Rep., 2019.[2] F. M. Vieira, P. S. Moura, and A. T. de Almeida, “Energy storage systemfor self-consumption of photovoltaic energy in residential zero energybuildings,” Renewable Energy , vol. 103, pp. 308 – 320, 2017.[3] J. Mohammadi, “Distributed computational methods for energy manage-ment in smart grids,” Ph.D. dissertation, CMU, 2016.[4] J. Delgado, R. Faria, P. Moura, and A. Almeida, “Impacts of plug-in electric vehicles in the portuguese electrical grid,” TransportationResearch D: Transport and Environment , vol. 62, pp. 372–385, 2018.[5] M. K. Gray and W. G. Morsi, “A novel transactive energy frameworkfor prosumers with battery storage and electric vehicles,” in , Oct 2018, pp. 1–6.[6] S. Behboodi, D. P. Chassin, C. Crawford, and N. Djilali, “Electric vehicleparticipation in transactive power systems using real-time retail prices,”in , Jan 2016, pp. 2400–2407.[7] Z. Liu, Q. Wu, K. Ma, M. Shahidehpour, Y. Xue, and S. Huang,“Two-stage optimal scheduling of electric vehicle charging based ontransactive control,” IEEE Transactions on Smart Grid , vol. 10, no. 3,pp. 2948–2958, May 2019.[8] S. R. Bender, M. Niemeyer, M. R. Weimar, and T. D. Hardy, “Con-siderations for commercial building participation in a transactive energysystem,” in , 2019, pp. 1–5.[9] H. Hao, C. D. Corbin, K. Kalsi, and R. G. Pratt, “Transactive controlof commercial buildings for demand response,” IEEE Transactions onPower Systems , vol. 32, no. 1, pp. 774–783, 2017.[10] S. Ramdaspalli, M. Pipattanasomporn, M. Kuzlu, and S. Rahman,“Transactive control for efficient operation of commercial buildings,” in , 2016, pp. 1–5.[11] E. Nefedov, S. Sierla, and V. Vyatkin, “Internet of energy approachfor sustainable use of electric vehicles as energy storage of prosumerbuildings,” Energies , vol. 11, no. 8, 2018.[12] Y. Kuang, Y. Chen, M. Hu, and D. Yang, “Influence analysis of driverbehavior and building category on economic performance of electricvehicle to grid and building integration,” Applied Energy , vol. 207, pp.427 – 437, 2017.[13] M. A. Quddus, O. Shahvari, M. Marufuzzaman, J. M. Usher, andR. Jaradat, “A collaborative energy sharing optimization model amongelectric vehicle charging stations, commercial buildings, and powergrid,” Applied Energy , vol. 229, pp. 841 – 857, 2018.[14] P. Moura, G. Yu, S. Sarkar, and J. Mohammadi, “Linking parking andelectricity values to unlock potentials of electric vehicles in portuguesebuildings,” in , 2020.[15] Z. Liu, Q. Wu, M. Shahidehpour, C. Li, S. Huang, and W. Wei, “Trans-active real-time electric vehicle charging management for commercialbuildings with pv on-site generation,” IEEE Transactions on Smart Grid ,vol. 10, no. 5, pp. 4939–4950, Sep. 2019.[16] P. Moura, G. Yu, and J. Mohammadi, “Management of electric vehiclesas flexibility resource for optimized integration of renewable energy withlarge buildings,” in , 2020.[17] P. Fonseca, P. Moura, H. Jorge, and A. T. de Almeida, “Sustainabilityin university campus: options for achieving nearly zero energy goals,”