Inter national J our nal of Electrical and Computer Engineering (IJECE) V ol. 11, No. 1, February 2020, pp. 182 200 ISSN: 2088-8708, DOI: 10.11591/ijece.v11i1.pp182-200 r 182 P o wer system operation considering detailed modelling of ener gy storage systems Ser gio Cantillo, Ricardo Mor eno Ener gy and Mechanical Department, Uni v ersidad Aut ´ onoma de Occidente, Colombia Article Inf o Article history: Recei v ed Apr 13, 2020 Re vised Apr 14, 2020 Accepted May 28, 2020 K eyw ords: Ener gy storage systems Generation dispatch Optimal po wer flo w Rene w ables sources ABSTRA CT The po wer system operation considering ener gy storage syst ems (ESS) and rene w- able po wer represents a challenge. In a 24-hour economic dispatch, the generation resources are dispatched to meet demand requirements consi dering netw ork restric- tions. The uncertainty and unpredictability associated with rene w able resources and storage systems represents challenges for po wer system operation due to operational and e conomical restrictions. This paper de v eloped a detailed formulation to model ener gy storage systems (ESS) and rene w able sources for po wer system operation in a DCOPF approach considering a 24-hour period. The model is formulated and e v alu- ated with tw o dif ferent po wer systems (i.e. 5-b us and IEEE modified 24-b us systems). W ind a v ai lability patterns and scenarios are used to assess the ESS performance un- der dif ferent operational circumstances. W ith r e g ar d to the systems proposed, there are scenarios in or der to e v aluate ESS performance. In one of them, the increase in capacity did not represent significant sa vings or performance for the system, while in the other it w as quite the opposite especially during peak load periods. This is an open access article under the CC BY -SA license . Corresponding A uthor: Ricardo Moreno, Ener gy and Mechanical Department, Uni v ersidad Aut ´ onoma de Occidente, Calle 25 # 115-85, Cali, Colombia. Email: rmoreno@uao.edu.co 1. INTR ODUCTION No w adays, the generation portfolio of electricity in po wer systems is more di v ersified than some years ago by the inte gration of rene w able resources [1]. En vironmental concerns are pushing the inte gration of technologies to produce electricity with rene w able resources [2]. As a result, there is an increasing to spur in v estments in order to diminish the con v entional fos sil fuel-based po wer generation [3–5]. Consequently , the international ener gy agenc y (IEA) reports that rene w able ener gy sources ha v e increased at an a v erage annual rate of 2.0 % from 1990 [6]. Gro wth is lar gely due to solar PV (37.4 %) and wind po wer (23.4 %) [6]. The i nherent features of this type of resources as uncertainty and v ariability impact po wer system operation [7–10]. In this conte xt, po wer systems require strate gies to inte grate such intermittent resources with fle xibility to meet the demand requirements [11]. The ener gy storage systems (ESS) represent a technology to store rene w able ener gy according to their a v ailability during the day (i.e., there are hi gh quantities of electricity from PV systems at noon). The ESS can absorb ener gy when generation e xceeds the load especially when this surplus come from rene w able sources and supply this ener gy to the grid during load peak hours [12, 13]. Thus, the ESS pro vides fle xibility under the inte gration of rene w able resources gi v en that the po wer dispatch can be settled to a desirable supply profile [14, 15]. J ournal homepage: http://ijece .iaescor e .com Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 183 The inte gration of ener gy storage systems (ESS) represent a challenge for the operation of po wer systems from dif ferent perspecti v es . The quality and reliability can be compromised due to mis use, misplacing or bad sizing of ESS [16]. Nonetheless, other challenges for po wer system operation are recognized, such as performance and safety (vie wed from its constituent materials, interconnections [17, 18], and service life), the distrib uted generation impacts in the po wer system coherenc y [19], the re gulatory en vironment, the in v estment costs, and the industry acceptance [20]. These issues can occur because system operation in v olv es decisions in dif ferent time frames (since minutes to days) including weather -dependent rene w able units scheduling and their reserv es [21] (e.g. wind [22]) as well as considering other relat ed v ariables. Ho we v er , the ESS mathematical modeling and its inte gration to po wer systems is a challenge with great impact and importance. The optimal po wer flo w (OPF) is used widely by po wer systems operators to dispatch economi cally the generation resources according to operational and economical restrictions [23]. From this perspecti v e, the po wer system operation requires a detailed modeling of storage systems in order to be included in the OPF mathematical formulati on. There are se v eral reasons for including these ener gy storage models in the economic dispatch. One of them is the more ef ficient inte gration of rene w able ener gy sources, since these de vices contrib ute to diminish the ef fects of the stochastic nature of these sources [24]. Also, the ESS contrib ute to maintain the stability in the po wer system operation, due to t he y restrict the fluctuation of instantaneous po wer coming mostly from rene w able sources [25, 26]. Lik e wise, the y allo w a more ef ficient economic dispatch since these de vices pro vide fle xibility that reduces the amount of po wer coming from more e xpensi v e sources (i.e., the y deli v er when there is a lack of ener gy , and store when there is a surplus), being cheaper and with less w aste [27]. Ho we v er , the inte gration of ESS’ s into an OPF model introduces inter alia, time interdependence. That is to say the ESS can char ge in periods of high wind or lo w demand (i.e. is absorbing po wer from the grid), and dischar ge in periods of lo w wind a v ailability or load peak (i.e is injecting po wer to the grid). This choice depends on the char ge status (i.e. SoC) at the pre vious time interv al and their respecti v e ef ficienc y . Also, technical and economical conditions are required to a v oid une xpected situations as char ging and dischar ging simultaneously . In other w ords, this situation implies that ESS w ould be paid for char ging and dischar ging at once [11]. Among others, the dual feature of absorbing and generating po wer requires a precise modelling for po wer system operation. This paper proposes a detailed formulation to include ESS in the optimal po wer flo w with mult i- ple generation sources to pro vide a 24-hour dispatching to meet demand requirements. Since ener gy storage systems could be defined as a generator and load due to the dual feature and also are time-correlated as men- tioned abo v e. The proposed formulation determines the optimal outputs for all g e neration portfolio as well as ESS char ging/dischar ging schedules seen through its SoC, all of them under dif ferent operation conditions and scenarios. The paper is or g anized as follo ws. The problem description and formulation are presented in Section 2. In Section 3, the 5-b us and IEEE 24-b us modified systems and their parameters are described. Then, the proposed procedure is tested using the systems described abo v e. At the end of this section, the results are analyzed and discussed. Section 4 pro vides some concluding remarks about this topic. - Literature re vie w The optimal po wer flo w for dispatching generation resources including rene w able sources has been widely discussed. The DC multi-period optimal po wer flo w (DCOPF) formulation ha v e been e xtended to include the v ariable nature of rene w able po wer generation, elements such as uncertainty in electricity demand and wi nd a v ailability [28–31]. Also, in some w orks other features such as branches and generation constraints are e xplicitly included in the formulation such as presented in [32–34]. Other authors ha v e made comparisons and anal ysis between this approach and con v entional methods without these v ariables [35]. On the other hand, some w orks emplo ys heuristic approaches including det erministic and stochastic methods (e.g Montecarlo simulation) to solv e the optimal dispatching [36–41]. Se v eral studies [42–45] ha v e researched the inte gration of intermittent wind po wer using a probabil is- tic approach. In order to pro vide better tools for the construction of gene ration scenarios and stochastic dispatch models [46–48]. Consequently , optimal po wer flo w has also been used with ESS in order to assess the po wer system operation fle xibility [49], due to these units can absorb ener gy in case of e xcessi v e generation or lo w electricity prices, mitig ating the uncertainty in the rene w able sources. Also in this research topic, studies such [50, 51] ha v e found other issues such as inclusion of ESS in distrib uted generation (DG) and RES with their respecti v e modelling and sizing. P ower system oper ation considering detailed modelling of ... (Ser gio Cantillo) Evaluation Warning : The document was created with Spire.PDF for Python.
184 r ISSN: 2088-8708 Other studies [11, 52–54] propose approaches in the economic dispatch using multi-period OPF due to specific challenges to the traditional OPF such as the modeling of char ge/dischar ge of ESS, or a specific ESS technology featuring [55]. Other studies ha v e included more v ariables in order to bring the problem closer to a more precise conte xt such as [56, 57] using po wer losses constraints on the transmission branches to e v aluate dif ferent generation scenarios. On the other hand, in [58] adds an en vironmental approach, modeling the social cost using v ariables such as emission generation in order to optimize the total production costs, using as little as possible the thermal generati on, without ne glecting the reliability in the system, all of this cases w orking under a DC approach. 2. DC-B ASED OPTIMAL PO WER FLO W WITH ESS This secti on includes t h e notation and the mathemati cal formulation for the mul tiperiod DCOPF dis- patching model including the ESS modeling. This model also includes thermal and wind po wer generation. 2.1. Notation g Thermal generation unit. i; j Netw ork b uses connected by transmission branches. t T ime period (hours). c , d Char ging/Dischar ging ef ficienc y of the ESS units. G Number of thermal generation units. L Number of netw ork branches. T T ime period in the operating horizon, in this case 24 hour . N Number of netw ork b uses. V W L W ind po wer w aste v alue ($/MWh). C ch ; C dch ESS Char ging/Dischar ging mar ginal cost ($/MWh). X ij Branch reactance connecting the i -b us to j . (p.u) b g Fuel cost coef ficient of thermal units ($). P max g , P min g Maximum/Minimum po wer generation thresholds of the thermal unit g (MW). P L max ij Maximum po wer flo w boundaries of branch connecting the i -b us to j (MW). P ch max , P ch min Maximum/Minimum char ge po wer limits for the ESS unit connected on the i -b us (MW). P dch max , P dch min Maximum/Minimum dischar ge po wer limits for the ESS connected on the i -b us (MW). C S max , C S min Maximum/Minimum ener gy stored (MWh). D i;t Electric po wer load in the i -b us at time t . Av w ind t W ind turbine a v ailability on the i -b us at time t (MW). C w ind t W ind turbine capacity connected on the i -b us (MW). R up g , R dow n g Ramp-up/do wn thresholds of thermal generation unit g (MW/h). P L ij ;t Acti v e po wer flo w from the i -b us to j -b us at time t (MW). P Gen i;t Acti v e po wer generated by thermal unit g at time t (MW). P w ind i;t Acti v e po wer of wind turbine connected to i -b us at time t (MW). P w l i;t Curtailed po wer of wind turbine connected to the i -b us at time t (MW). i;t Dual v ariable that denote Locational Mar ginal Price in the i -b us at time t ($/MWh). F obj 24-hour T otal operating costs ($). i;t V oltage angle of the i -b us at time t (rad). C S i;t Ener gy stored in the i -b us at time t (MWh). P ch i;t , P dch i;t Po wer Char ged/dischar ged to/from ESS connected to the i -b us at time t (MW). 2.2. F ormulation The formulation is e xpressed as optimization problem to address a mi nimum total operating cost associated with producing electricity to meet the demand for a 24-hour period described by (1). In (2) indicates the total cost of ener gy production with g thermal units during an interv al of time T . In (3) refers to the production costs associated with not taking full adv antage of the source of wind generation a v ailable during this same interv al of time. In (4) represents a condition that requires that the ESS are not char ged and dischar ged simultaneously , this pre v ents the payment of an ESS for char ging and dischar ging simultaneously [11, 29, 59], situation that cannot occur . Int J Elec & Comp Eng, V ol. 11, No. 1, February 2020 : 182 200 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 185 F obj = C G + W L + C E S S (1) C G = T X t =1 G X g =1 b g P Gen g (2) W L = T X t =1 N X i =1 V W L P w l i;t (3) C E S S = T X t =1 N X i =1 ( C dch P dch i;t C ch P ch i;t ) (4) The restrictions for the dispatching model are gi v en by the po wer flo w equations. This paper uses the DC approach to include po wer flo w calculations. The po wer flo w balance is gi v en by (5). The po wer flo wing on each line is gi v en by (6). The po wer flo w restrictions are gi v en by the boundaries in the (7). G X g =1 P Gen g ;t + P w ind i;t D i;t P ch i;t + P dch i;t = L X j =1 P L ij ;t (5) P L ij ;t = 1 X ij ( i;t j ;t ) (6) P L max ij ;t P L ij ;t P L max ij ;t (7) The dual v ariable associated to (5) correspond to the locational mar ginal price (LMP) of each b us hourly . On the other hand, the restrictions for thermal generation units are defined in (8), (9), and (10), where (8) corresponds to the operational range of thermal generators. On the other hand, (9) and (10) indicates the maximum up and do wn ramps limits that each of the thermal gener ators can perform from one hour to the ne xt. P min g ;t P Gen g ;t P max g ;t (8) P Gen g ;t P Gen g ;t 1 R up g (9) P Gen g ;t 1 P Gen g ;t R dow n g (10) The ener gy le v el (i.e. State of char ge) of ESS were defined per unit in the i -b us at time interv al t , depends on the dif ference between the ESS char ged and dischar ged po wer with their respecti v e operating ef ficiencies, as defined in (11). The maximum and minimum limits of ESS char ge/dischar ge, and ESS Capacity were defined in (12), (13) and (14) respecti v ely . C S i;t C S i;t 1 = c P ch i;t P dch i;t d (11) P ch i;min P ch i;t P ch i;max (12) P dch i;min P dch i;t P dch i;max (13) C S i;min C S i;t C S i;max (14) The restrictions for wind generation (i.e. wind po wer loss) are defined in (15). The e xpression cor - responds to the reduction of use of potentially a v ailable wind ener gy . In (16) describes the minimum and maximum po wer range that a wind generator can produce, considering placing and wind a v ailability . P ower system oper ation considering detailed modelling of ... (Ser gio Cantillo) Evaluation Warning : The document was created with Spire.PDF for Python.
186 r ISSN: 2088-8708 P w l i;t = Av w ind t C w ind i P w ind i;t (15) 0 P w ind i;t Av w ind t C w ind i (16) 3. RESUL T AND DISCUSSION In order to test this approach to study a wide range of applications, initially , a small case and then a modified IEEE standard case are used to illustrat e the ESS modelling in a multi-period dispatching and sho w their performance according to dif ferent operational situations. This section pro vides a comprehensi v e e xplanation of each case and the corresponding analysis to observ e ESS performance during a 24-hour period. All simulations were completed by a computer (PC) running W indo ws R with an Intel R Core I5+ 8300H processor @2.3 GHz with 12.00 GB RAM, using Gurobi R Solv er (8.1.1) [60] under the JuMP 0.20.1 Julia platform [61]. 3.1. Load cur v e description The daily load curv es used for the 5-b us (orange) and 24-b us (blue) po wer systems are plotted in Figure 1. The load curv es present four (4) decreasing trend bands with its lo west point at hour 4 (i.e. 787.1 MW and 1950.6 MW respect i v ely), and three (3) increasing trend bands with a load peak at hour 20 (i.e. 1150 MW and 2850 MW respecti v ely). Figure 1. Load curv e pattern for po wer systems testing 3.2. W ind a v ailability pr ofiles Three (3) wind profiles are construc ted to e v aluate the ESS performance during the operation of both po wer systems considering wind po wer a v ailability (i.e. lo w , moderate, and high) as sho wn in Figure 2. The simulation results of both po wer systems, such as the the thermal generators scheduling and the ESS performance as well as their respecti v e analysis can be found in the follo wing subsections. Figure 2. W ind a v ailability profiles used for po wer systems testing Int J Elec & Comp Eng, V ol. 11, No. 1, February 2020 : 182 200 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 187 3.3. ESS perf ormance in a 5-Bus system 3.3.1. Case description The one-line diagram for a 5-Bus system is sho wn in Figure 3. This system includes thermal genera- tion, wind generation and st o r age. The thermal unit parameters are listed in T able 1, modifying the information from [49]. The load is distrib uted in 4 b uses. Figure 3. One-line diagram of modified 5-b us po wer system T able 1. Thermal generation info for the 5-b us po wer system Gen Bus P min g (MW) P max g (MW) Mar ginal Cost (MW) R up g (MW/h) R dow n g (MW/h) 1 1 0 140 17 20 20 2 1 0 170 18 25 25 3 3 0 360 20 30 30 4 5 0 490 21 35 35 T able 2 lists the netw ork grid information such as reactance and rating in MV A (i.e po wer line con- straints), all of them modified from [49]. T able 2. Branch info for the 5-b us test system Fr om T o X ij (p.u) Rating (MV A) 1 2 0.0281 400 1 5 0.0064 400 2 3 0.0108 400 4 5 0.0297 240 The 5-b us test system includes a wind po wer plant connected to the b us 4. The wind po wer generation site and capacity is listed in T able 3. Also, this system includes an ESS connected in the b us 2. In other w ords, the ESS is not on the same b us as the wind po wer plant. The ESS parameters considered are ESS capacity , char ging and dischar ging ef ficienc y , and operating v alues. Such features are listed in T able 4. T able 3. W ind po wer generation info for the 5-b us system Gen Bus C w ind i;t (MW) 1 4 240 T able 4. ESS info for 5-b us system. ESS Bus Capacity (MW) c (%) d (%) C S i;min (%) C S i;max (%) 1 2 50 90 90 10 90 P ower system oper ation considering detailed modelling of ... (Ser gio Cantillo) Evaluation Warning : The document was created with Spire.PDF for Python.
188 r ISSN: 2088-8708 3.3.2. Results The simulations use the 5-b us po wer system with the parameters gi v en before (i.e. load curv e and wind profiles) to e xplore and e v aluate dif ferent operational situations. Initially , the performance of the po wer system w as e v aluated according to gradual increases of the ESS capacity , starting from its base capacity (i.e. 25 MW steps, starting at 50 MW up to 200 MW). The analysis highlights changes in the thermal generation scheduling and ESS performance during the 24-hour period. The ESS performance (i.e State of Char ge (SoC)) during a 24-hour period is sho wn in Figure 4. Lik e wise, the ESS char ging interv als occurs at hours 3 t o 7, 16 to 18, and 23 to 24. A one ESS dischar ging interv al occurs in the load peak v alue (hours 19 to 21). The ESS is char ged in v alle y hours (lo w demand) and dischar ged at load peak hours (i.e. time shifting ef fect and transmission curtai lment reduction) as e xpected. On the other hand, the ESS performance sho ws a g ap when its capacity reaches 150 MWh and the wind a v ailability impro v es (i.e. moderate and high a v ailability). This finding is presented in hours where there is no char ging or dischar ging beha vior (i.e. hours 5 to 16). Figure 4. (a) Comparison of ESS performance between the lo w-wind a v ailability , (b) the moderate wind a v ailability , (c) and high-wind a v ailability The description of the ESS performance leads to the analysis that the of ESS installed capacity could be o v ersized due to wind a v ailability . This could happen in lo w-wind a v ailability due to wind turbines and Int J Elec & Comp Eng, V ol. 11, No. 1, February 2020 : 182 200 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 189 thermal units w ouldn’ t ha v e enough po wer to contrib ute meeting demand and char ge the ESS at the same time, unless the wind a v ailability increases. Therefore, tw o or more ESS with dif ferent capacities could ha v e similar SoC v alues where the higher capacities are underutilized. Since it w ould not ha v e complete char ging c ycles (e.g 200 MW and 175 MW ESS capacities in all wind patterns) thus decreasing its life c ycle, and represents a smaller reducti on in operating costs. This analysis sho ws that e v en in s mall po wer systems, features such as the ESS capacity must be analyzed technically and economically in a strict w ay . On the other hand, the dif ferent thermal generation schedules according to the ESS capacity increas es during a 24-hour period are sho wn in Figure 5. Similar performances to the proposed demand curv e are presented especially in the lo w-a v ailability wind pattern. Nonetheless, such performances mo v ed a w ay as wind a v ailability increases (i.e. moderate and high a v ailability patterns) as in the case of ESS performance. Furthermore, it ca n be appreciated dif ferences in thermal scheduling between ESS capacities on v alle y hours (i.e hours 2 to 6, and hours 17 and 18) of the load curv e for all wind patterns. Also, another dif fer ence between scheduling is presented at the peak of the load curv e (i.e hours 20 and 21). Figure 5. (a) Comparison of thermal scheduling between the lo w-wind a v ailability , (b), the moderate wind a v ailability , (c), and high-wind a v ailability Additionally , the thermal units dispatching under dif ferent wind a v ailability patterns sho ws that wind a v ailability determines the dispatching of therm al units. The proposed system has a limited generation portfolio P ower system oper ation considering detailed modelling of ... (Ser gio Cantillo) Evaluation Warning : The document was created with Spire.PDF for Python.
190 r ISSN: 2088-8708 with a strong dependence on thermal units and lo w wind po wer par ticipation. this f actor e xplains the closeness between thermal scheduling and the load curv e specially under lo w wind a v ailability patterns and ho w similar beha viour is maintai ned re g ardless of wind a v ailability . In the same w ay , the thermal unit scheduling between the highest proposed ESS capacities (e.g. 175 and 200 MW) are similar . This finding pro v ed the misuse of ESS from a certain capacity and wind a v ailability as mentioned abo v e. Lik e wise, this issue represents a non-impro v ement of the po wer system performance as wel l as a ne gligible reduction of thermal generation compared with increases in the ESS capacity . Moreo v er , The ESS performance seen from its SoC during a 24-hour period is sho wn in Fi gure 6. Unlik e the pre vious scenario, It sho ws no dif ferences for some of the proposed a v ailability patterns. Since in lo w a v ailability pattern, the ESS presents the same beha vior re g ardless of the increase in mar ginal cost (i.e from 1.0 to 2.0 times). In all cases, char ging and dischar ging patterns are presented depending on the respecti v e wind pattern beha vior . Ho we v er , there is a consistent unloading pattern during peak hours (i.e. from hours 19 to 21). This represents the correct ESS modeling and operation since it deli v ered po wer during the load peak as e xpected. Figure 6. (a) Comparison of ESS performance with re g ard to the mar ginal cost increasing, between the lo w-wind a v ailability , (b) the moderate wind a v ailability , (c) and high-wind a v ailability Also, the ESS performance sho ws a direct influence by wind a v ailabili ty due to the f act that the ESS Int J Elec & Comp Eng, V ol. 11, No. 1, February 2020 : 182 200 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Elec & Comp Eng ISSN: 2088-8708 r 191 finds some operation fle xibility by increasing wind a v ailability . Thus, in lo w wind a v ailability the storage system in most of the time tends to char ge until load peak hours re g ardless of the mar ginal cost of the thermal units, while in moderate or high wind a v ailability depending on the mar ginal cost dif ferent beha viors can be presented (that is to say po wer amounts and char ge or dischar ge decisions) of the ESS. Nonetheless, although dif ferent performances according to the mar ginal fuel cost are presented, in general terms the ESS performed in a similar w ay . In other matters, the thermal units dispatch according to wind a v ailability and compared to the load curv e is sho wn in Figure 7. It sho ws similar beha viors between the thermal scheduling and the load curv e for all wind a v ailability profiles, in some cases (i.e. hours 1 to 7 in lo w-wind a v ailability) the load curv e and the thermal units dispatch ha v e matched. Thus, the thermal unit dispatch also sho ws fe w changes by increasing the mar ginal cost to the proposed v alue. These changes were presented when load f alls (i.e. hours 3 to 7 and hours 19 to 24) and e xist high wind a v ailability , as sho wn in c). F or all other wind a v ailability patterns, the same thermal units po wer dispatch w as presented. Figure 7. (a) Comparison of thermal unit scheduling with re g ard to the mar ginal cost increasing, between the lo w-wind a v ailability , (b) the moderate wind a v ailability , (c) and high-wind a v ailability Furthermore, the the wind a v ailability ef fect on the po wer system is e vident since the dif ference be- tween load and thermal units po wer is greater (i.e. dif ferences between lo w , moderate and high wind a v ailability P ower system oper ation considering detailed modelling of ... (Ser gio Cantillo) Evaluation Warning : The document was created with Spire.PDF for Python.