Centralized flexibility services for distribution system operators through distributed flexible resources

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/337565938 Centralized flexibility services for distribution system operators through distributed flexible resources Conference Paper · October 2019 CITATION 1 READS 116 4 authors: Some of the authors of this publication are also working on these related projects: Vehicle to Microgrid (V2M) View project New solar control tensed devices for workshops about sustainable school architecture View project Sara Barja-Martinez Universitat Politècnica de Catalunya 7 PUBLICATIONS4 CITATIONS SEE PROFILE Pol Olivella-Rosell Universitat Politècnica de Catalunya 33 PUBLICATIONS317 CITATIONS SEE PROFILE Pau Lloret-Gallego Universitat Politècnica de Catalunya 22 PUBLICATIONS161 CITATIONS SEE PROFILE Roberto Villafafila-Robles Universitat Politècnica de Catalunya 84 PUBLICATIONS1,590 CITATIONS SEE PROFILE All content following this page was uploaded by Sara Barja-Martinez on 27 November 2019. The user has requested enhancement of the downloaded file. Centralized flexibility services for distribution system operators through distributed flexible resources Sara Barja-Martinez1, Pol Olivella-Rosell1, Pau Lloret-Gallego1and Roberto Villafafila-Robles1 Centre d’Innovaci´o Tecnol`ogica en Convertidors Est`atics i Accionaments (CITCEA-UPC), Departament d’Enginyeria El`ectrica, ETS d’Enginyeria Industrial de Barcelona, Universitat Polit`ecnica de Catalunya, Avinguda Diagonal, 647, Pl. 2, 08028 Barcelona, Spain, [email protected] Abstract. Under the context of smart grids within smart cities, increasing distributed generation, consumer empowerment and emerging flexibility services, distribution system operators could benefit by activating flexibility in distribution grids to avoid deploying new infrastructures and grid overloading. The solution offered by this work is an energy management system algorithm capable of activating flexibility behind the prosumer main meter during constrained periods. Therefore, the distribution system operator could compensate grid congestion during high consumption or production periods and increase their renewable generation hosting capacity by using behind-the-meter flexibility during peak production periods. Keywords: smart grids, smart cities, distribution grid, flexibility, energy management system, centralized optimization 1 Introduction The presence of more intermittent distributed generation and the empowerment of consumers are forcing the power system to evolve and adapt its operation to these changes. Past electrical system was mainly based on centralized, dispatchable and predictable generation that provided flexibility at transmission level to the electrical system to balance generation and demand. However, the increasing installation of distributed renewable generation is transforming the generation side in a more variable and intermittent source of energy. At the same time, the European Commission has presented a package of measures to ensure that consumers are active and central players on the energy markets of the future [1]. In this sense, the use of flexibility from the demand side can boost the involvement of prosumers in the energy system and make them a valuable asset in the electrical market. The proper management of available flexibility, both in generation and demand side, can help to compensate the lack of certainty of renewable sources. 2 Sara Barja-Martinez et al. In addition, electric vehicles (EV) and heat pumps have a strategic role in reducing greenhouse gas emissions and they are a key component of the transition to a low carbon economy [2]. However, its widespread use is increasing the demand of electricity, which may cause the need to upgrade the electricity infrastructure. The introduction of flexibility services can also be used as a more efficient alternative to reinforce the distribution grid, reducing or postponing infrastructure investment needs [3]. The use of flexibility for congestion management in the distribution grid is currently being widely investigated and there are some undergoing initiatives trying to standardize and provide common understanding of flexibility usage in the distribution grid. As an example, Universal Smart Energy Framework (USEF) Foundation created a detailed framework to provide an integral market design for the trading of flexible energy use [4]. However, optimization strategies are not covered as they can be different for each flexibility operator based on its own requirements and characteristics. Regarding optimization strategies, [5] proposes a method to employ the flexibility service from EV and heat pumps for real-time congestion management through an optimal power flow. In contrast, authors in [6] presented an optimization framework for the use of customers flexibility aggregation participating in the wholesale power market and the regulation capacity market. Moreover, in [7] an optimization problem is formulated considering battery degradation cost and using a decomposed solution approach with the alternating direction method of multipliers (ADMM) instead of commonly adopted centralised optimization to reduce the computational burden and time, and then reduce scalability limitations. This paper presents a centralized energy management system algorithm that provides flexibility from prosumers to distribution system operators (DSOs) during constrained periods in order to avoid grid congestion or other related grid failures. The suggested approach has been developed under the INVADE project [8], which aims to design a flexibility management system using batteries that supports the distribution grid and electricity market while coping with grid limitations, high penetration of renewable energy and EV. The main contribution of this paper is the development of a robust algorithm capable of activating the maximum flexibility available to meet the DSO flexibility request (FR) at minimum cost behind the prosumer main meter when needed to avoid grid congestion during high consumption or production periods at distribution level. The remainder of the paper is organized as follows: Section 2 describes the developed framework. The mathematical formulation problem is outlined in Section 3. The case study and its results are presented in Section 4. Ultimately, conclusions are presented in Section 5. 2 System description The optimization problem description and the architecture implemented is based on the INVADE H2020 [8] project. The result of this project will be an integrated platform enabling flexible management algorithms. It will be applied to public Centralized flexibility services for distribution system operators 3 and private EV charging stations, households and mid-size customers to offer flexibility services to DSOs, BRPs and prosumers. The three main actors involved in the present work and what role each one of them play is described below: Distribution system operator : requests and purchases flexibility to the aggregator in order to avoid grid congestion and gives the corresponding financial compensation to the aggregator. Aggregator : receives flexibility requests from the DSO and flexibility offers from the prosumers that are part of its portfolio. It is responsible for activating the flexibility requested by the DSO at minimum cost. Prosumer : is the flexibility provider. Each prosumer aims to minimize its electricity bill by optimizing the used of batteries and photo-voltaic (PV) generation, but this optimized baseline consumption can be altered if the DSO needs to avoid a failure in the distribution grid in a certain period. The aggregator will economically reward the prosumer for modifying his optimized baseline. The relationship and how these 3 main actors interrelate among each others is shown in Fig. 1. Fig. 1. Local flexibility market agents outline Decisions are taken centrally. The aggregator will try to satisfy the DSO flexibility request at the lowest possible cost within its portfolio. This approach has a two way communication, which means that local data is available as an input to the optimization algorithm and the central system has direct control on the local flexible electric devices [9]. 2.1 Flexibility Services Flexibility services can be classified in function of the flexibility customer. The three main flexibility customers are listed in Table 1 along with what kind of 4 Sara Barja-Martinez et al. Flexibility customer Flexibility Service Distribution system operator Congestion management Voltage / Reactive power control Controlled islanding Balance responsible party Day-ahead portfolio optimization Self-balancing portfolio optimization Prosumer Hourly tariff optimization kW max control Self-balancing Table 1. Main flexibility customers and their flexibility services. flexibility services they can demand. In order to describe the flexibility services, the INVADE project [10] and article [11] are used as reference. The present study focuses on the DSO and prosumer flexibility service. Prosumers aim to minimize their electricity bill, while the DSO requests then flexibility needed to operate properly the distribution grid, within the safe operation zone. 2.2 DSO flexibility requests The DSO flexibility requests are the minimum required amount of active energy variation with respect to the aggregated prosumer baseline optimization to avoid grid overloading. Negative flexibility request values mean decreasing generation or increasing demand while positive flexibility request is defined as increasing generation or decreasing demand. Table 2 summarizes these definitions. FR <0 FR >0 ↓generation ↑consumption ↑generation ↓consumption Charge batteries Discharge batteries Table 2. Description of the DSO flexibility requests The proposed flexibility algorithm flow chart is described in Fig. 2 and it is based on [7]. The algorithm starts by minimizing each prosumer electricity bill using their flexibility devices: distributed batteries and PV generation. The optimization result is the aggregated baseline optimized, which is the sum of each of the optimized consumption prosumer sites. The following step is to check whether the portfolio has enough flexibility to meet the DSO requests: the aggregator executes the aggregated level flexibility offer (ALFO) optimization problem. In case the flexibility requested could not be activated, it would deliver as much as possible. The aggregator sends to the DSO a flexibility offer and if Centralized flexibility services for distribution system operators 5 the DSO accepts it, the aggregated level flexibility management (ALFM) optimization problem is then carried out, which will provide the flexibility asked to the DSO at minimum cost. Fig. 2. DSO flexibility service flowchart 2.3 Flexibility Sources Flexibility in the distribution grid can come from different distributed energy sources, mainly grouped by three types: loads, storage units, and renewable generation. All these flexible sources that can provide the amount of flexibility requested by the DSO are listed and described below: Demand-side response: prosumers are provided with a financial incentive to turn down or turn off non-essential processes at times of peak demand or high energy prices, depending on what you want to minimize or maximize, helping the grid to balance supply and demand without the need for additional generation to be used. Energy storage systems: electricity systems face an increased need for flexibility and a fundamental pillar in terms of flexibility are batteries, whose main potential is to help to deal with the high volatility of distributed renewable energy resources. Energy can be stored when there is a surplus of renewable energy generation. This energy can then be used at a time when its needed. Distributed Energy Resources: they are electric generation units located within the electric distribution system at or near the end user. Electricity is generated locally, minimizing transportation losses. 6 Sara Barja-Martinez et al. 3 Mathematical problem formulation This problem is executed in three main steps: first, an individual optimization of each of the prosumers is carried out. The aggregated prosumer baseline optimized is used as an input to the ALFO optimization problem. Once the flexibility offer is calculated by the aggregator and accepted by the DSO, the ALFM optimization problem is carried out and it will provide the flexibility service to the DSO. 3.1 Individual prosumer flexibility service optimization problem Prosumer objective function and constraints are defined as follows: The prosumer optimization function (1a) aims to reduce the electricity bill by minimizing the amount of energy purchased to the grid χbuy t, taking into account the revenues for injecting electricity to the grid χsell t, maximizing the generation of renewable energy resources and minimizing the flexibility cost ζflex t, which is the minimum amount of money that the prosumer is willing to save in order to activate a flexibility source (1b). Equation (1c) represents the internal energy balance behind each prosumer smart meter: the total electricity imported from the grid, the optimized production from PV generation units ψgen,r tand the energy discharged by batteries σdis tmust be equal to the electricity exported to the grid, the consumption from inflexible load units Winflex tand the energy charged in batteries σch tfor each period of time. Binary variables δbuy tand δsell tare introduced in equation (1d) in order to ensure that it is not possible to sell and buy electricity in the same period. They are set to 1 if the customer is buying (importing) or selling (exporting); else 0. The amount of electricity bought (1e) and sold (1f) in each period of time must be less or equal to the maximum energy export capacity of each site per period, according to the terms stipulated in the retail contract. min χ, ζ X t∈T (Pretail,buy tχbuy tPV AT t−Pretail,sell tχsell t+ζflex t) (1a) s.t. ζflex t=Pgen,r t(Wgen,r t−ψgen,r t) + Pb,ch tσch t+Pb,dis tσdis t,(1b) ψgen,r t+σdis t+χbuy t=χsell t+σch t+Winflex t,(1c) δbuy t+δsell t≤1,(1d) χbuy t≤δbuy tXmax,import,(1e) χsell t≤δsell tXmax,export (1f) Centralized flexibility services for distribution system operators 7 3.2 DSO flexibility service optimization problem The aggregator has to ensure that there is enough flexibility available in his portfolio to meet the DSO flexibility request. The DSO purchases this available flexibility and gives the corresponding economic compensation to the prosumer through the aggregator. Once the flexibility offer sent by the aggregator is accepted by the DSO, the ALFM optimization problem is be executed. The objective function (2a) is to minimize the aggregator operational cost of meeting DSO flexibility request. Wbaseline,opt p,t is the aggregated baseline consumption after the individual prosumer optimization. Constraints (2b) (2c) ensure that the activated amount of flexibility is less or equal to the positive and negative FR, respectively. Constraints (2d) and (2e) avoid the rebound effect, which can cause new load peaks before or after the FR activation. min χ, ζ X t∈T X p∈P (Pretail,buy tχbuy p,t PV AT t−Pretail,sell tχsell p,t +ζflex p,t ) (2a) s.t. χbuy p,t −χsell p,t ≤Wbaseline,opt p,t −FRt∀FRt>0,(2b) χbuy p,t −χsell p,t ≥Wbaseline,opt p,t −FRt∀FRt<0,(2c) χbuy p,t ≤max(Wbaseline,opt p,t ),(2d) χsell p,t ≤max(Wbaseline,opt p,t ) (2e) 3.3 Distributed Flexible Resources constraints Energy Storage System constraints Distributed storage units can provide flexibility to the electrical grid by charging or discharging batteries to meet a flexibility request made by the DSO in a given period of time. The behaviour of the battery is then formulated. The variable σsoc tin (3) indicates the current battery state of charge (SOC). With the aim to represent a more accurate battery model, the mathematical formulation has into account efficiency factors for storing ηch and delivering electricity ηdis. Energy storage units can meet both, negative and positive DSO flexibility requests by charging σch tor discharging σdis tthe batteries, respectively. Both are variables in this problem. σsoc t=σsoc t−1+σch t·ηch −σdis t ηdis ∀t∈T(3) In order to preserve and extend the battery life-time, σsoc tmust be between a minimum Omin and a maximum Omax energy limit value (4): Omin ≤σsoc t≤Omax ∀t∈T(4) Equations (5)(6) limit the maximum energy charged Qch and discharged Qdis by battery per period . σch t≤Qch Nhour ∀t∈T(5) 8 Sara Barja-Martinez et al. σdis t≤Qdis Nhour ∀t∈T(6) Equation (7) makes sure that the energy charged σch tis linearly decreased.Sch b is the threshold in charging process. The same happens with the discharging energy σdis t(8).The lower threshold to limit the energy output is Sdis b. σch t≤−Qch 1−Sch (σsoc t Omax −1) ∀t∈T(7) σdis t≤Qdis Sdis σsoc t Omax ∀t∈T(8) The total battery degradation cost ζbat is taken into account (9). Pb,ch is the degradation price for charging 1 kWh. The discharging degradation cost has been already included in the charging cost. ζbat =X t∈T Pb,ch ·σch t∀t∈T(9) Photo-voltaic reducible generation constraints The optimized PV scheduled production ψgen,r tmust be between 0 and the PV baseline electricity generation Wgen,r t. The price for reducing the PV generation is set high to maximize the renewable generation. 0≤ψgen,r t≤Wgen,r t∀t∈T(10) The total cost for reducing the PV generation is given by ζgen,r.Pgen,r is the price for disconnecting a PV generation unit. ζgen,r =X t∈T Pgen,r ·(Wgen,r t−ψgen,r t)∀t∈T(11) 4 Case Study A case study where the aggregator provides a flexibility service to the DSO and prosumer in the Spanish energy market is proposed. The aggregator controls the flexible energy sources of its portfolio, which is formed by 31 prosumers located in Austin, Texas. Real load consumption and PV generation data from different households have been provided by DataPort Inc. Street [12]. The present case study covers a planning horizon of 3 days, divided into 15-minutes time intervals and starting at April 1st of 2019 at 00:00h. All the distributed storage units in the aggregator’s portfolio begin and end at half their maximum capacity. The Spanish tariff market is applied in the present optimization problem and the elecricity tariff chosen for buying electricity from the grid is the PVPC (Precio Voluntario para el Peque˜no Consumidor), because