1
Optimal management tool for micro grids with a high penetration of renewable energy sources Dr.-Ing. Pio Lombardi1, Dipl-Ing. Xiubei Ge2, Dipl.-Ing. Tatiana Sokolnikova3, Prof. Dr.-Ing. Zbigniew Styczynski4
Abstract--Some Renewable Energy Sources (RES) such as wind and solar will become the backbone of the future power system. However, because their power generation is not programmable, the integration of the generated electric power into the electric grid may be a not easy task. Increasing the generation capacity of the plants based on RES also increases the effort to feed the electric power into the grid. Micro grids are power systems which may be disconnected from the main power network and may work as autonomous systems. Generally, the power (electric and thermal) generated within the micro grid is locally managed. The Energy Management System (EMS) is the heart of a micro grid since it optimally controls the power generated from the conventional power plants as well as the power stored inside the Energy Storage Systems (ESS) based on the power generated by RES and on the power demanded. A tool to optimally manage a micro grid with a high penetration of RES has been developed. The tool simulates the behavior of an EMS which optimally controls the conventional generators and the ESS. The criteria which drive the EMS are the costs and the generation by RES. Through the developed tool, the conventional generators are optimally dispatched to minimize the generation costs, while the ESS is managed to maximize usage of the power generated by RES. Index Terms--Autonomous micro grid, energy management system, energy storage systems, mixed integer linear programming, renewable energy sources, smart grid.
I. INTRODUCTION
T
He Renewable Energy Sources(RES) have been increasingly developed over the last decades, driven by environmental issue, the critical situation of fossil fuel and incentive policies in many countries. However, the main challenge of utilizing such sources comes from the nonprogrammable and fluctuating power generation from RES. Conventional power grids were usually constructed without considering the integration of such dispersed power generation, along with the increasing integration of renewable energy. To overcome this difficulty, the concept of micro grid was raised. Micro grids could ease the negative influence of integrating renewable power generation into the main power grid and are also applicable for further extended amounts and types of
renewable sources [1], [2]. However, the generation of solar and wind sources could not stably supply load demand due to its uncertain nature, which is especially severe for girds with small capacities. Therefore, some stable power sources such as diesel generators and energy storages are usually required in autonomous micro grids to obtain the balance between demand and supply [2]. An Energy Management System (EMS) is the heart of a micro grid since it optimally controls the power generated from the conventional power plants as well as the power stored inside the Energy Storage Systems (ESS), depending on the power generated by RES and on the power demanded. This study describes the development of a tool which simulates the behavior of an EMS. The tool optimally controls the conventional generators and the energy storage system. The criteria which drive the EMS are the costs and the generation by RES. Through the developed tool, the conventional generators are optimally dispatched to minimize the generation costs through the use of mixed integer linear programming [3], while the ESS is managed to maximize usage of the power generated by the RES. II. MICRO GRIDS AND EMS MODELING A. Micro grid modeling A micro grid that is disconnected from the main power network was modeled. It consists of several conventional generators, a wind farm, a PV plant and an energy storage system (Fig. 1). An EMS optimally controls the charge/discharge schedule of the ESS and the production of the conventional generators according to their generation costs, the weather conditions and the State of Charge (SOC) of the storage system.
1
Fraunhofer Institute for Factory Operation and Automation IFF, Magdeburg, Germany,
[email protected] 2 Otto-von-Guericke University Magdeburg, Germany,
[email protected],
[email protected] 3 Irkutsk State Technical University, Irkutsk, Russia
[email protected]
Fig. 1 Scheme of a micro grid
2
B. Flow chart of EMS With reference to the Fig. 2, the EMS receives real-time input of the load, wind power, solar power and storage SOC. Then it determines the storage charging or discharging power, the unit commitment of conventional generators, and the stability state of the micro grid, i.e. if the system is confronting a blackout or if the power generated by RES is not able to be fed in. The EMS modus operandi is depicted as a flow chart in Fig. 2 and Fig. 3, where i is the index time point, ‘time’ refers to the time span of each calculation step, and ‘eff’ refers to the charge/discharge efficiency of the ESS. The EMS tool firstly sets the parameters of the ESS, including the maximum charging and discharging power (stated in the Fig 2 as Storage rated power), the charging and discharging efficiency and the storage capacity. Beside them, the EMS tool considers as a first step the parameters of the conventional generators including the rated power and the linearized fuel cost functions. The method of the linearized fuel cost functions is described in Section B. Secondly, the EMS tool at the begin of every step elaborates the real-time wind power; the solar power and the SOC. As first result, the EMS estimates the maximum charging or discharging power (depicted as Pch or Pdisch), the deficit or the surplus power (depicted as Pblackout and Psurplus), and finally the unit commitment of the conventional generators (depicted as Pgen), which is detailed in Section B.
Fig. 3 Flow chart of an EMS (part 2)
Fig. 4 Flow chart of an EMS (part 3)
C. Unit commitment of conventional generators One of the most important functions of an EMS is to optimize the operation cost of a micro grid by minimizing the fuel cost of conventional generators. A quadratic cost function has been used which relates the fuel cost to the generated power (1). (1) 𝐶(𝑃) = 𝑎 ∙ 𝑃2 + 𝑏 ∙ 𝑃 + 𝑐 Since the mixed integer linear programming algorithm has been chosen to optimally schedule the generators, the cost functions and constraints have been linearized. Thus, the first step for an EMS to optimize generator usage is to linearize the quadratic cost functions of generators. Fig.5 shows how the tool breaks the cost function into segments.
Fig. 5 Linearization of quadratic cost functions Fig. 2 Flow chart of an EMS (part 1)
The objective function as well as the constraints of the optimization problem are shown in (2) and (3), respectively. 4
𝑂𝐹1 = 𝑚𝑖𝑛 ∑ 𝐶𝑖 (𝑢𝑖 , 𝑃𝑖 ) 𝑖=1
(2)
3 4
{
TABLE I: GENERATOR DATA
∑ 𝑃𝑖 (𝑡) = 𝐿𝑜𝑎𝑑(𝑡) − 𝑃𝑤𝑖𝑛𝑑 (𝑡) − 𝑃𝑝𝑣 (𝑡) ± 𝑃𝑐ℎ/𝑑𝑖𝑠𝑐ℎ (𝑡) 𝑖=1
(3)
Generator
Pmax [MW]
Pmin [MW]
Fuel cost coefficients
Start up costs [€]
𝑢𝑖 ∙ 𝑃𝑖𝑚𝑖𝑛 ≤ 𝑃𝑖 ≤ 𝑢𝑖 ∙ 𝑃𝑖𝑚𝑎𝑥
Where Ci are the generation costs, ui is a binary variable [0 or 1], (1 if the generator is set up) Pi is the produced power from the generator i, t is the time step Load is the demanded load in for the hour t, Pwind is the power generated during the hour t by the wind farm, Ppv is the power generated during the hour t by the PV, Pch/disch is the power charged and discharged to/from the ESS during the hour t. This power is positive when the ESS is charged and negative when it is discharged. III. CASE STUDY A. Problem formulation As case study, a microgrid with a maximal electric load of 12 MW has been analyzed. Two scenarios have been considered: in the first one a wind farm and a PV plant supply the electrical power to the microgrid, while in the second scenario, besides the wind farm and the PV plant, the power is supplied by four conventional generators and an energy storage system. In the two scenarios the wind park and the photovoltaic plant have both an installed power of 8 MW. The conventional generators are all able to generate up to 1 MW of electric power, while the storage power and the storage capacity are respectively 3 MW and 5 hours. Fig 6 shows profiles of the load and RES generation in a typical summer week. If the system does not use any conventional generator and any energy storage system (first scenario) then 99.4 MWh of RES is unfed-in the microgrid and 493.4 MWh of load is not satisfied. By considering the costs for the unfed-in RES as 25 €cent/kWh [4] and the costs for the unsupplied load as 20 €cent/kWh [5] then the system operation costs for the analyzed week are 123.5 k€. The question now is: with four conventional generators as shown in Table I and an ESS whose parameters are shown in Table II, how could operational costs be saved?
1 1 1 1
1 2 3 4
0 0 0 0
a [€/MW2h] 0.01 0.023 0.026 0.024
b [€/MWh] 30 42 32 97
28 28 28 28
TABLE II: STORAGE PARAMETERS Rated power(MW)
Capacity(h)
3
5
Charge/discharge efficiency 90%
B. Simulation results Simulation results for the second scenario show that the unfed-in RES in the week is reduced to 49.2MWh and the lost load is reduced to 74.1MWh, whereby 380.1MWh of the load is supplied by conventional generators. The operation costs are 84.9k€ per week. By comparing the two scenarios the use of conventional generators and energy storage system contribute to reduce the operation costs of 31%. ESS charge and discharge schedule and the change of SOC are presented in Fig.7 and Fig.8, respectively. Change of lost load, unfed-in RES and the dispatch of conventional generators are shown in Fig.9 and Fig.10, respectively. At the hour 20, from Fig.6, one can see that the RES generation is insufficient for the load; Fig.7-Fig.10 show that the EMS fully loads all the conventional generators and discharges the ESS till it is empty, but still a lost load of 3.8 MW could not be compensated. At the hour 100, an unfed-in RES of 6.9 MW happens since the ESS has already been fully charged.
Fig. 7 ESS charge/discharge power in one week
Fig. 8 Change of SOC in ESS in one week
Fig. 6 Profiles of load, wind generation and solar generation in one week
c [€/h] 109 97 109 100
4
VI. BIOGRAPHIES
Fig. 9 Lost load and unfed-in RES in one week
Pio Lombardi studied mechanical engineering at the Politecnico di Bari, Italy. He graduated in 2006 at the same university with the degree M.Sc. He joined the Chair of Electric Power Networks and Renewable Energy Sources at the Otto-vonGuericke University Magdeburg, Germany as a research engineer in 2006. At the same university he received his PhD. In 2011 he joined the Process and Plant Engineering of Fraunhofer Institute for Factory Operation and Automation IFF. His primary field of interest includes modeling, simulation and optimization of Smart Grids. He is a member of the Baikal project research group.
Xiubei Ge studied mechanical engineering in Beijing University of Posts and Telecommunication, China, and received her B.Sc degree in 2010. She joined a doubledegree program operated by Otto-von-Guericke University of Magdeburg, Germany, and Wroclaw University of Technology, Poland, graduating in 2012 with an M.Sc degree in the field of renewable energy systems and electrical engineering. She has worked as a research assistant in the Faculty of Electrical Engineering and Information Technology in Otto-vonGuericke University of Magdeburg since March 2012.
Fig. 10 Unit commitment of conventional generators in one week
IV. CONCLUSIONS A tool for simulating the behavior of an EMS was developed. The EMS optimally controls the ESS and the production of conventional generators. Scheduling the charging and discharging of the ESS is mainly driven by RES generation, and the dispatch of conventional generators is decided based on minimizing fuel costs. Simulation results show that the EMS does help micro grids with cost savings, and the calculation speed is relatively high, which means that the development of EMS in practical cases is feasible. V. REFERENCES [1]
[2]
[3] [4] [5]
P. Lombardi, P. Vasquez, Z. Styczynski, “Optimised autonomous power system” in Proc. 2009 Cigre IEEE PES Joint Symposium Calgary, 29 July 2009. P. Lombardi, T.Sokolnikowa, Z. Styczynski, “Optimal storage capacity within an autonomous micro grid with a high penetration of renewable energy cources”, IEEE PES ISGT Berlin Jizhong Zhu, “Optimization of power system operation”, IEEE,2009 CRA International “Assessment of the Value of the Customer Reliability (VRC)”, August 2002. K. K. Kariuki, R. N. Allan, “Applications of Customers Outage Costs in System Planning, Design and Operation” IEEE Proceeding – Generation, Transmission and Distribution 143, 305-312, 1996
Tatyana V. Sokolnikova graduated in 1985 with M.Sc. from the Irkutsk State Technical University (ISTU) in Hydrogeology. Between 1985 and 2005 she was a leading planning engineer in the Planning Institute Irkutsk. In 2008, she completed her master's degree in Smart Grid technology at the ISTU and now is working in the scope of the Bajkal Projekt on her Ph.D.-. Her research interests are related to the planning and optimization of autonomous Smart Grids, taking into account the role of energy storages.
Zbigniew A. Styczynski (SM ‘01) received his PhD in EE at the Technical University of Wroclaw. He worked at the Technical University of Stuttgart, Germany and 1999 he became Chair of Electric Power Networks and Renewable Energy Sources of the Faculty of Electrical Engineering and Information Technology at the Otto-vonGuericke University, Magdeburg, Germany. Since 2006 he has also been the president of the Centre of the Renewable Energy Saxonia- Anhalt, Germany. His special field of interest includes modelling and simulation of the electric power networks systems, renewable, and optimization problems. He is the author of more than 150 scientific papers, a senior member of IEEE PES, a member of CIGRE SC C6, VDE ETG and IBN and a fellow of the Conrad Adenauer Foundation. In 2011, together with the Irkutsk State Technical University (Project Baikal), he won the Super Grant of the Russian Federation and is leading a research group at ISTU in the scope of Smart Grids.
