Energy Management of a Hybrid Photovoltaic-Wind System with Battery Storage: a Case Report

This work presents a case report related to the management and the monitoring of a hybrid photovoltaic-wind system with battery energy storage, installed at the administrative offices building of the municipality of Valderice (Italy) within the framework of the Italy-Tunisia ENPI cooperation project DE.DU.ENER.T. Le Développement durable dans la production energétique dans le territoire. The paper describes the hybrid system and briefly reports the monitoring data for a whole year, comparing the real production with the expected one and evaluating some performance indexes of the system. The performance indexes are very simple and have been defined only with the purpose of showing the advantages of distributed generation. Then, two different control logics for the battery energy storage systems are compared in order to define the most suitable management of the local energy resources, in presence of different Time-of-Use electricity tariffs. In particular, the two logics are compared by varying the difference between the electricity prices in peak hours and in off-peak hours and the rate between the electricity consumption of the building and the battery energy storage’s capacity.


INTRODUCTION
The development of sustainable technologies for electrical systems and of new renewable energy sources (RES) based generators, together with the deregulation of the electricity market, has led to a growing penetration of distributed generation (DG), building automation and control (BAC) systems and ICT applications and components, thus transforming the electrical distribution system.In this scenario, the issue of the optimal management of the available electrical [1] and thermal [2] energy resources and loads [3] also at end-user level is getting more and more attention.
At the same time, in the last few years, Li-ion batteries have become common components of commercial photovoltaic systems [4], opening the way to new possibilities for the optimal exploitation of the energy produced by such systems, in particular, in the presence of net-metering policies or residential time-of-use tariffs [5].As an example of the works in literature on this topic, in [6] the behavior and efficiency of a low-cost isolated photovoltaic system for typical rural houses near Luena in Angola are examined, while in [7] the authors present an economic optimization of battery operation for two different applications in residential buildings under different dynamic tariff structures.In [8] the performance of a household battery energy storage system with a Li-ion battery pack and a single-phase converter are studied.In [9] a new methodology to enable high penetration of photovoltaic (PV) systems in low voltage (LV) grids by using shared battery storage and variable tariffs is proposed.In [10], a technical economical analysis is performed for determining the cost optimal configuration for a PV plant with batteries considering the share of self-consumption, the degree of autarky, grid feed-in and supply as well as various battery system parameters.
Moreover, thanks to Li-ion batteries, Demand Response (DR) policies [11] can experience new possibilities.In [12] the authors describe a distributed control method for residential BESS coupled with PV plants for using customer owned storage units for solving the over-voltage issues caused by high PV penetration.In [13] an overview of some recent initiatives for exploitation of local storage systems and DR as a solution for achieving a stable operation of renewable energy sources-based microgrids.In [14] the issue of voltage quality improvements using PV systems is discussed.
For exploring the opportunities given by RES generators coupled with innovative storage systems, two different hybrid Photovoltaic-Wind systems with storage were built in Valderice (Italy) and in Borj-Cédria (Tunisia) within the framework of the international cooperation project DE.DU.ENER.T-Le Développement durable dans la production energétique dans le territoire [15].The systems comprise: • a PV generator; • a micro-wind generator; • a battery energy storage system (BESS).In literature some works deal with hybrid RES systems and are mainly focused on autonomous power supply in case of grid shortage [16], operation of grid-off systems [17], applications related to rural electrification and remote community [18] or power quality improvements [19].
The current work, starting from the data obtained from one year of monitoring of the Valderice installation, analyzes the behavior of the hybrid system evaluating some performance indexes and shows a comparison between two different algorithms for the management of the BESS.The two BESS management algorithms have been presented in [20] and are tested here, taking into account various electricity tariffs, suitable for the considered end-user.In the following, the comparison between the algorithms is done considering only the daily economic savings, having assumed that the system is already in operation.Therefore, no considerations on the capital investment cost or on the economical viability of the investment are provided.
The rest of the paper is structured as follows: • the two BESS management algorithms are described and four performance indexes are defined for comparing them; • the hybrid system is introduced and the production data in the period March 2016-February 2017 are shown; • a comparison of the algorithms in various cases is presented changing the electricity tariff; • finally, the conclusion of the work are given.

METHODS
In this section, the BESS management control logics are presented together with three indexes for their comparison.

BESS management control logics
Two different management control logics, named System Led and Market Led, for combined RES-BESS generators connected to the LV utility grid are described in the following.The control logics have been implemented with the aim of: • increasing the self-consumption quota in order to minimize the exchange of energy between the end-user and the grid; • reducing the yearly energy bill, fully exploiting the advantages of TOU tariffs.

System Led
According to the System Led control logic, the Energy Management System (EMS) installed at the end-user's facility acquires: • the energy generated by the PV system (Ep1); • the energy generated by the wind system (Ep2); • the energy consumption of the building (Ec); • the BESS state of charge (Est).On the basis of these data, the EMS command the combined RES-BESS system according to the flow chart represented in Figure 1.
In the case that the total production exceeds the building's energy consumption and the state of charge (SoC) of the batteries is lower than the maximum SoC (Est_max), the energy in excess is stored in the batteries, otherwise is injected into the grid.
On the contrary, in the case that the energy production is lower than the energy consumption of the building, if the SoC is higher than the minimum SoC (Est_min) the batteries supply the building; otherwise, the required energy is taken from the grid.

Market Led
As formulated, the System led algorithm is not able to take into account the effect of Time of Use (TOU) tariffs and consequently, it cannot be used for an optimal economic exploitation of the BESS.
For this reason, a second control logic has been defined, named Market led algorithm.This control logic is based on the variations of the electricity price during the day (and the week).The Market led control logic has been defined considering the three pricing periods for Italian end-users [21][22] reported in Table 1.
The end-users' EMS, equipped with an internal clock and calendar, is set for recognizing the pricing period at every moment of the day.
In this case, the Market led control logic operates according to the flow chart in Figure 2. When the building's electricity consumption is lower than the electricity production from the generating unit, different situations can happen: • during pricing period F1, the energy in excess is injected into the grid because the selling price is the highest and the end-user can achieve the maximum revenue; • during pricing period F2, the production in excess is stored in the batteries.In this case, the charge of the batteries rises till Est_max/2.When this limit is exceeded, the production in excess is injected into the grid; • during pricing period F3, the energy in excess is stored in the batteries again.In this case the charge of the batteries rises till Est_max.When this limit is reached, the energy in excess is injected into the grid.Off-peak price period: all remainder hours of the week and holidays.
On the contrary, when the building's consumptions exceed the generated electricity: • during pricing period F3, the grid supplies the building because the electricity price is the lowest and the cost for the end-user is minimum; • during pricing period F2, the energy is taken from the batteries till the SoC is over Est_max/2 and when the SoC gets below this value, the energy is taken from the grid; • during pricing period F1, the energy is taken from the batteries till the SoC is over Est_max; when the SoC gets below this value the energy is taken from the grid.

Performance indexes
Three easy to evaluate and to understand performance indexes have been defined for comparing the System and Market led strategies.Indeed, their definition has been led by the purpose of providing a clear and simple set of indicators that can be easily understood even by non-technicians, in order to promote the diffusion of the culture of RES-based system among a broader public.
The three indexes are: • Energy Saving index EnS; • Economic Saving index EcS; • CO2 reduction index CO2R.

Energy Saving index
The Energy Saving index is defined as: where Ea is the electric energy demand of the building in a month and Eb is the energy produced by the hybrid system in the same month.

Economic Saving index
The Economic Saving index is defined as: where Ea1, Ea2 and Ea3 are the electric energy demand of the building in the three price periods F1, F2 and F3 in a month, Eb1, Eb2 and Eb3 are the energy produced by the hybrid system in the same month for the three price periods and c1, c2 and c3 are the electricity prices in the three price periods.

CO2 reduction index
The CO2 reduction index is defined as: where Eb is the total energy produced by the hybrid system in one year and Ce is the conversion coefficient from electric energy to Tons of CO2.

RESULTS
In this section, the two algorithms are applied to the management of the BESS of the hybrid PV-Wind system built in Valderice.Before presenting the results of the comparison, the system is described.

System description
The hybrid system is installed at the administrative offices building of the Municipality of Valderice, located close to Trapani (geographical coordinates 38.02N 12.36E) in Italy (Figure 3).
The building is characterized by four floors and has a flat roof and external fixtures made of anodised aluminium and single glass without shutters.The covering of the building has a sufficiently regular shape and has a useful surface of about 400 m 2 , with parapet walls and an elevator fence (Figure 4).
The building is supplied by the Low Voltage (LV) utility grid through a three-phase connection with available power of 20 kW, is equipped with a centralized gas boiler and has single air-conditioning units only in some rooms.The study of the solar radiation map and wind speed allows the characterization of the site chosen.The building insists on a geographic area characterized by solar photovoltaic productivity of about 1650 kWh/kWp per year and wind productivity of about 2000 kWh/kW per year.
After performing the energy characterization of the building, the rationalization and energy efficiency measures and the areas available for the installation of the plants have been identified and, on the basis of the available budget, the optimal size of the systems micro-generation and electrical storage system have been calculated and the necessary components and devices have been chosen.
Thus, the prototype is composed of: • a 8 kWp PV system in polycrystalline silicon, consisting of 32 x 250Wp modules divided into two strings; • a 1 kW vertical-axis wind turbine (VAWT); • an electrical storage with capacity of 2 kWh (lithium-ion batteries) integrated into one of the two inverters of the photovoltaic system; • a solar-thermal system with flat collectors and a 250 liters tank for hot water storage serving 4 bathrooms and replacing electric water heaters; • an indoor lighting controller for the offices, in order to contain the relative electrical consumption.The following figures 5-8 show the main components of the prototype.The hybrid power plant benefits from a net-metering contract.Tables from 2 to 4 report the technical data of the PV and wind generators.As shown in Table 3, the PV system comprises two 4 kW inverters instead of only one 8 kW device.This choice has been made with the aim of comparing the performance of the two inverters in the presence and in the absence of storage.

Results of the measurement campaign
Figure 9 shows the actual daily energy produced by the PV system compared with the expected values calculated by simulations.Thanks to the energy meters installed at the point of common coupling with the distribution grid and at the output terminals of the hybrid generators, both the energy produced by the hybrid system and the energy demand of the building have been measured for one year.By using these data, the four indexes defined above are calculated.Table 5 reports the results of the calculation of EnS and EcS.The values in Table 5 show that due to the installation of the hybrid system the building has yearly energy savings of about 68% and yearly economic savings of about 64%.
Finally, considering the Italian energy mix, the coefficient for converting the electricity consumptions into avoided emissions of CO2 gas is equal to 494.30 gCO2/kWh [23].This implies that CO2R=1.97Tons of CO2/year.

Comparison of the control logics
A MATLAB code has been written for simulating the behaviour of the RES-BESS system of Valderice, receiving as input the real monitored production data reported above.
With regard to the consumption of the building, the following occupancy schedule has been assumed for calculating it: from 8:00 to 14:00 from Monday to Friday, from 16:00 to 18:00 on Tuesday and Thursday.The offices have been assumed closed on Saturday and Sunday with no electricity consumption on these days.Starting from the real average yearly consumption of the building of about 32000 kWh, and from the occupancy schedule above, the following daily consumptions have been assumed for the simulations: 124 kWh for MON, WED, FRY, 170 kWh for TUE, THU, and 0 kWh for SAT, SUN.
The daily load profiles built starting from the above assumptions and used in the simulations are reported in Figure 0. The comparison between the two control logics has been done considering one week as an example.Figure 11 reports the daily production of the hybrid system in a week of May 2016.The red curve has been obtained as average of the real consumptions from Monday to Friday.
The comparison is done considering only the power produced by the PV system given that the wind system does not supply the batteries.Nevertheless, the control logics are also applicable to the case of wind turbine connected to the BESS and the results of the study are not influenced by the actual configuration of the system under examination.
Six cases are considered: • CASE 1: the electricity price is equal to 0.11 €/kWh, 0.0969 €/kWh and 0.077 €/kW in the three pricing periods and the daily consumptions are derived by Figure 10; • CASE 2: the electricity price is equal to 0.11 €/kWh, 0.0969 €/kWh and 0.077 €/kW in the three pricing periods and the daily consumptions are 50% of those considered in CASE 1; • CASE 3: the electricity price is equal to 0.22 €/kWh, 0.0969 €/kWh and 0.077 €/kW in the three pricing periods and the daily consumptions are derived by Figure 10; • CASE 4: the electricity price is equal to 0.22 €/kWh, 0.0969 €/kWh and 0.077 €/kW in the three pricing periods and the daily consumptions are 50% of those considered in CASE 3; • CASE 5: the electricity price is equal to 0.39 €/kWh, 0.0969 €/kWh and 0.077 €/kW in the three pricing periods and the daily consumptions are derived by Figure 10; • CASE 6: the electricity price is equal to 0.39 €/kWh, 0.0969 €/kWh and 0.077 €/kW in the three pricing periods and the daily consumptions are 50% of those considered in CASE 6. Table6 reports the revenues and the costs due to the sale of electricity between the building and the utility, evaluated for the System Led and the Market Led algorithms.The electricity price has been assumed equal to 0.11 €/kWh, 0.0969 €/kWh and 0.077 €/kWh for pricing periods F1, F2 and F3, respectively (CASE 1).
The percentage economic saving of the Market Led algorithm with respect to the System Led algorithm is 0.56%.Therefore, in the considered case the new algorithm is not able to give a significant advantage with respect to the classical one.
Nevertheless, the performance of the Market Led algorithm depends on the rate between the production and the demand of the building.Considering the same system with consumptions equal to 50% of those represented in Figure 0, the results in Table 7 are obtained (CASE 2).The percentage economic saving of the Market Led algorithm with respect to the System Led algorithm is about 10%.
It is important also to investigate how the Market Led algorithm behaves in presence of high differences between F1 and F23 electricity prices.Therefore, simulations have been carried out with the two different consumption profiles considered above and imposing the following electricity prices: 0.22 €/kWh, 0.0969 €/kWh and 0.077 €/kWh for pricing periods F1, F2 and F3, respectively.The elaboration gives the results reported in Table 8 (CASE 3) and Table9 (CASE 4).The percentage economic saving of the Market Led algorithm with respect to the System Led algorithm is about 1.2% and 18% for the third and fourth considered cases, respectively.
Finally, other two extreme cases are presented (CASE 5 and CASE 6 reported in Table 10 and Table 11).The following prices are considered: F1 0.39 €/kWh, F2 0.0969 €/kWh, F3 0.077 €/kWh.Finally, Table 12 reports the comparison of the results for the six examined cases.
In particular, Table 3 shows that: • in case of low difference between peak and off-peak prices, the Market led algorithm does not provide any significant benefit for the end-user with respect to the System led algorithms.Moreover, increasing the size of the storage with respect to the electricity consumptions, poorly increase producer's revenue; • in case of high difference between peak and off-peak prices, the Market led algorithm assures higher revenues to the producer and the benefits increase when consumptions are lower (or, alternatively, if the storage capacity increases).

CONCLUSION
The paper has presented the management of a hybrid PV-wind system with batteries installed at the administrative office of Valderice in the framework of the ENPI DE.DU.ENER.T research project.Four indexes have been presented for evaluating the performance of the system.Moreover three different algorithms for managing the BESSs integrated with the hybrid generator have been proposed and discussed.
Simulations show that, in particular, the Market Led algorithm in certain specific condition but not always is able to generate economic saving for the owner of the hybrid system with respect of the System Led algorithm.It has been shown that the savings depend on the rate between production and building electricity consumption and on the difference between the electricity prices in the peak price period and in the other pricing periods.
Market Led algorithm is particularly suitable for public administration office buildings, characterized by reduced electricity consumptions in the weekend.Indeed, this allows completely recharging the BESS and restarting the work cycle at the beginning of every week.
It is worth nothing that the results of the comparison between the System and the Market led algorithm are valid in the case that the peak price period corresponds to the highest production period of the PV system (daylight hours).In some countries, for example in islands not supplied by the mainland with high values of generation from PV systems, electricity prices are structured so as to encourage the consumption during day hours, presenting higher values during the evening and in the night.
Therefore, the results of this study must be carefully evaluated and limited to the situations that can be assumed similar to those considered in the simulations.

Figure 1 .
Figure 1.Block diagram representing the System Led control logic.

Figure 2 .
Figure 2. Block diagram representing the Market Led control logic.

Figure 3 .
Figure 3.The administrative offices buildings of the Municipality of Valderice.

Figure 4 .
Figure 4. Rooftop of the building.

Figure 9 .
Figure 9. Average daily electricity production profile from PV: experimental and simulated data.

Figure 10 .
Figure 10.Daily load profiles of the building (red line for MON, WED, FRI, green line for TUE, THU).

Figure 11 .
Figure 11.Daily production profile of the hybrid system (weekdays: red line; Saturday: blue line; Sunday: green line).

Table 2 .
Technical data of the PV modules.

Table 3 .
Technical data of the PV inverters.

Table 4 .
Technical data of the wind generator.

Table 5 .
Calculation of EnS and EcS Indexes.

Table 32 .
Comparison of the Results.