Optimal Power Sharing between Photovoltaic Generators, Wind Turbines, Storage and Grid to Feed Tertiary Sector Users

A paramount technical drawback of photovoltaic (PV) and wind power systems is the intermittency of energy production, which results in problems of stability of the electricity grid and power quality issues [1] [2] [3]. The effects of the separate installation of PV or wind power systems are well-known in literature [4] [5] [6]. In different countries, such as Denmark for wind, renewables have become the main sources of power: non-programmable RES share is currently around 50% of the national electric consumption. In Germany and Spain, this share is already higher than 20%, while in Italy this goal will be reached in few years [7]. In every case, a full transformation of the power system will be even more necessary: infrastructure, policies and markets have to be improved. To compensate for the intermittency of PV and wind generators, electrochemical storages are easy to install and manage in whatever site with respect to hydroelectric pumping systems which require large reservoirs. The widespread utilization could address the power balance of local loads and distributed generators mitigating their negative effects on the grid. For example, the power surplus from PV near midday could be used later to feed local loads. Nevertheless, storage is currently expensive and cannot solve the problem of the weak seasonal correlation between low demand and high non-programmable RES generation and vice versa. Therefore, it is fundamental to know what are the acceptable amounts of grid-connected RES and storages capacities. In particular, the maximum capacities of non-programmable RES must be defined so that there is no necessity of grid upgrade. Such an upgrade of distribution transformers and lines is a consequence of high active powers from distributed generation in reverse flow on radial networks originally designed to feed purely passive loads. Taking into account the previous general remarks, in the present PhD dissertation, the optimal power sharing between PV generators, wind turbines, storage and grid to feed tertiary sector (telecommunication equipment and offices) users is determined taking into account some technical end economic constraints. Therefore, the electrical consumers in this analysis are the owners of the PV generators, wind turbines and storage systems. These three technologies are used to meet a substantial amount of the demand, while the remaining power is provided by the distribution grid. In this way, the consumers became prosumers. First, each renewable source (sun and/or wind) is investigated in terms of hours of availability to estimate the total time in which PV and/or wind power productions occur along the whole year. Then, the primary goal of the prosumers is assumed to be the achievement of the best match between power profiles of loads and power profiles of generators. Such a best match is obtained thanks to an appropriate procedure to design the sizes of generators and storages. In this procedure, power ratings of PV and wind generators and energy capacities of batteries are chosen to reach the highest levels of self-consumption and the lowest levels of power exchange with the grid according to the load profile. In every case, the selected sizing solutions are cost-effective (Net Present Value NPV>0) and cannot create problems to the grid management (overloads of distribution lines and transformers are avoided). In Figure 1 1, it is present the scheme of the simulated systems: the main components are PV and wind generators, storage and electronic converters. They supply aggregations of tertiary-sector users with a peak of consumption corresponding to tens of megawatt, composed of many offices buildings and electronic equipment’s. The core of the system is the DC bus, which connects all the renewables at the voltage imposed by the batteries. On the contrary, all the loads are in AC and can be fed by the grid, when there is no production and storage is empty. PV generators are connected to the DC bus by DC/DC converters operating as Maximum Power Point Trackers (MPPTs) to extract the maximum power. An inverter provides AC power to user loads: this converter is unidirectional, because the storage recharging is performed by PV and wind generators without the grid contribution. The reason is that storage is used to support renewables to feed loads and decrease grid-exchanges. Charge-discharge cycles performed to buy and sell energy at different prices are not allowed and batteries are recharged only when there is a surplus from wind and PV (with respect to the local loads). Simulations start from hourly load profiles and environmental parameters. Solar irradiance G, air temperature Ta and wind speed data u, with 1-minute time step, are measured with high accuracy instruments of meteorological stations. These stations are installed in five sites in Apulia (Southern Italy) and have a mutual maximum distance of 150 km. The meteorological data are inputs of energy production models (Figure 1 2): the sites are characterized by a similar annual solar radiation and high deviations in wind speed profiles. The first result is that the maximum levels of self-consumption, that can be reached according to the abovementioned constraints, are in the range 50÷61% of the energy consumption. A so high autonomy from the grid is achieved by installing high storage capacities: they permit to use more PV generators and wind turbines with negligible grid injections. If the goal is the maximization of the NPV, the costs of grid exchanges become the main driver: storage is expensive and to sell energy to the grid is less profitable than self-consumption. The best choice is to design the PV and wind generators to satisfy an amount of the total consumption; in this way, most of RES production is self-consumed and injections are low. In this case, in which storage is not used, the maximum levels of self-consumption result in the range 34÷41% of the load. At the end of the simulation part of my thesis, the data related to the five case studies are compared with the results of two additional simulations, in which loads and generators of all the five analysed sites are aggregated. In the first aggregation case, distributed storage is not present and capacity of PV and wind generators corresponds to the sum of the capacities of the five sites, previously calculated in case of maximization of NPV. The aggregation of generators and loads permits to increase the self-consumption from an average value of 37% up to 40% for the whole system (this result corresponds to that one of the best case study). In future works, this improvement could reach much higher values, if other different typology of loads (e.g. dwelling houses and factories) will be added in the aggregation. In the second aggregation case, a centralized storage is present and its capacity corresponds to the sum of the capacities of distributed storage calculated in the five independent sites in case of maximization of self-consumption. The capacity of PV and wind generators corresponds to the sum of the capacities of the five sites. The aggregation does not permit to achieve better results with respect to single cases, because the high installed capacity of storage already well manages surplus and deficit of energy. In addition to the above described simulation activity, during the whole PhD course, experimental work was performed on PV systems. Part of this experimental work was published in three journal papers [8] [9] [10] (I am co-author) and they are reported in Chapter 4. This experimental work was fundamental to study in details the operation and issues of PV systems. Thus, this work was significant for the achievement of simulation results described in the previous chapters. In the first part of the experimental work, it is defined a reasonable procedure, in terms of minimum type and number of tests and thus minimum duration (a few days), in order to identify the sources of poor performance and to solve or mitigate their negative effects. Such a procedure is based on experimental tests, partly on the PV-system site and partly in laboratory, and the suitable data processing, both before the experiments and after them [8]. The second part of experimental work is related to the electric characterization of the PV generator by tracking its I-V characteristic by connecting it to a capacitor. The curve tracer based on capacitive loads is used, because it is simpler, cheaper and scalable from module level to array level with respect to the use of a variable controlled load. Nevertheless, it is required a detailed sizing of the capacitive load to optimize the duration and the accuracy of the measurements. The practical setup of I–V curve tracers at module, string and array levels is addressed in this work for the main commercial technologies. Such tracers represent a tool to easily check the performance of PV systems at the beginning and during their operation, especially when poor performance occurs [9]. In Chapter 1, it is present a description of PV and wind power generation systems and electrochemical storage technologies. In Chapter 2, it is described the architecture of the simulated systems. The management of energy flows, the models of renewable generators, simulation constraints and sizing criteria are also described in detail in Chapter 2. In Chapter 3, inputs of the simulation and the energy and economic results of the different case studies are presented. Finally, the experimental work on PV system is reported in Chapter 4.