Self-Su ﬃ cient and Islanded-Oriented Design of a Reversible Solid Oxide Cell-Based Renewable Microgrid

: This article presents the development of a constrained optimization algorithm, whose scope is to support the preliminary design of a renewable microgrid, integrating solar panels and wind turbines with reversible solid oxide cells. The motivations behind this research activity lie in the increasing interest in renewable-based production and on-site storage of hydrogen, and its aim is to help this energy vector spread worldwide and in as many industrial and residential sectors as possible within a reasonably short timeframe. To this end, suitable models were developed by referring to the most relevant literature and by introducing some speciﬁc simplifying assumptions. Such an approach allowed the setting-up of a multi-variable constrained optimization task, whose outcomes correspond to the most techno-economic e ﬀ ective plant conﬁguration with respect to assigned design criteria. The optimum solution was particularly sought via the generalized reduced gradient method, aimed at determining renewable plants sizes under the constraint that the ﬁnal stored hydrogen level is brought back to the initial value after one year. The results highlight that an interesting payback time of about 10 years can be attained, while guaranteeing that the optimal conﬁguration holds promising resiliency and islanded-use capabilities (such as almost weekly self-su ﬃ ciency) via smart over-the-year charge-sustaining management of the designed hydrogen storage tank. In this way, it was possible to simultaneously address, via the speciﬁc optimization problem formulation, the interconnected needs of optimally designing system components in terms of installed power, and the proper management of the reversible solid oxide cell unit.


Introduction
The progressive increase in global energy requirements constantly pushes governments and involved researchers to find clean alternatives to traditional energy production methods. Damage to human health and the environment, increased greenhouse effect, lower availability, and higher prices are consequences of fossil fuel consumption that are in fact difficult to manage. The use of renewable energy sources (RES), which have the particularity of being clean and quickly regenerable, is now a viable solution due to related technological improvement. However, the progressive increase in the use of RES entails adopting adequate energy storage systems, so as to better adapt to the variability and uncertainty of power production. Currently, batteries are the most widespread; however, achieving the best compromise between performance and costs [1], limiting degradation [2], and avoiding overheating [3] still represent critical aspects. As for alternative storage devices, flywheel technology emerges as a suitable candidate, especially when high power density, fast response, and high efficiency to as useful guidelines, to be taken into consideration when developing component-level control strategies for the entire balance of plant (BoP), as well as when the aim is to carefully and cost-effectively select materials and types of key components such as photovoltaic panels, wind turbines, batteries, hydrogen tanks, etc. It is worth here remarking how in the literature some contributions dealing with aspects similar to the above ones are available. Raj and Chanana [19] proposed advanced fuel cost minimization-based management of an islanded microgrid comprising wind turbine, photovoltaic panels (PV), a microturbine, a diesel generator, and a battery bank. Their main objective was to optimize fuel consuming device scheduling for an already designed system configuration. Specific activities were developed in [20] to ensure stable power supply of RES microgrids in rural areas. The importance of system resiliency and energy security was suitably addressed in [21], via the development of a remote ZigBee-communications-based energy management tool. Dynamic stability operation was addressed once again by Luo et al. [6], this time including rSOC into a highly integrated renewable-based hybrid energy generator. As clarified later on, the main novelty of the current article with respect to the above contributions lies in the optimization task, which was proven suitable to simultaneously address optimal system sizing and proper islanded-oriented and resilient management of the rSOC system that the renewable microgrid is integrated with. Moreover, when compared with a previous contribution from the authors [22], the current paper focuses on the development of an rSOC microgrid for residential uses, thus referring to the typical electric load demand of households while suitably accounting for the impact of weather uncertainties (e.g., clouding effects, wind speed limits) on renewables power contribution. Moreover, substantial discussion is provided on the compromise to be found between self-sufficiency targets and hydrogen tank size exploitation levels.
The article is organized as follows. The characteristics of the selected case study are presented and discussed in Section 2, in view of its subsequent integration with the rSOC device. Afterwards, Section 3 describes the adopted modeling approach, emphasizing its simplicity to enhance the subsequent constrained optimization phase. The outcomes of the latter are presented and discussed in Section 4. Finally, the concluding remarks are given, aimed not only at confirming the economic viability of the proposed microgrid, but also at stressing the flexibility in meeting different loads, as well as the useful resilient features rSOC integration introduces.

Plant Description
The selected case study consists of a residential complex served by the following electric power sources: photovoltaic panels (PVs); vertical axis wind turbines (WT); hydrogen storage tank (HST); and an rSOC system. Figure 1 illustrates a simplified plant scheme, where the power demand coming from the residential complex is primarily met by the WT and PVs. Whenever the PV and WT yields overcome the above-mentioned electricity demand, the resulting excess power is converted into hydrogen by rSOCs (i.e., SOEC mode). The produced fuel gets stored in the HST for subsequent use, particularly when PV and WT contribution is lower than current demand. The inverters, which are necessary to perform DC-AC conversion, are also shown in the figure. The proposed rSOC-based renewable microgrid can be applied for residential distributed generation (as an islanded microgrid, particularly when public grid connection is missing), as well as to supply hydrogen to industrial processes and refueling stations. Figure 2 shows the average daily trajectory of electricity demand (P demand ). Aimed at guaranteeing appreciable islanded potential, rSOC-based microgrids must be designed so as to be self-sufficient. Therefore, beyond meeting load demand in cases of PV and WT energy shortage, the HST shall be designed in such a way as to provide appreciable self-sufficiency in the case of a long-lasting lack of other source contribution (e.g, due to system failures, relevant maintenance operation, and so on).  [23], including the reversible solid oxide (rSOC)based renewable microgrid, connection to the public grid and served residential complex. It is worth remarking here that all acronyms appearing in the above figure are explained in the nomenclature section at the end of the paper Figure 2. Electricity load demand for the entire residential complex, consisting of seven apartments. The illustrated trajectory was initially extracted from 24 and then adapted to suitably represent the average daily profile of dwelling electricity demand [24].
As for the photovoltaic panels, the choice fell on the panels described in [25], which are highperformance polycrystalline units, connected to the load network through a DC/AC inverter (see Figure 1). On the other hand, in order to have a lower cut-in wind speed, vertical axis technology was adopted for wind turbines [26]. Table 1 provides the main technical specifications for the assumed PV and WT modules, which are referred to when running the optimal sizing procedure, as discussed later on in Section 3.2). Differently from the RES, for the rSOC system the energy conversion performance, in both SOFC and SOEC operation, that correspond to maximum power operation were assumed by referring to the relevant contribution found in the literature [27].
For the electrical storage system, the most widely adopted performance index, i.e., rSOC roundtrip efficiency (RTE), is expressed as follows [28]:  [23], including the reversible solid oxide (rSOC)based renewable microgrid, connection to the public grid and served residential complex. It is worth remarking here that all acronyms appearing in the above figure are explained in the nomenclature section at the end of the paper Figure 2. Electricity load demand for the entire residential complex, consisting of seven apartments. The illustrated trajectory was initially extracted from 24 and then adapted to suitably represent the average daily profile of dwelling electricity demand [24].
As for the photovoltaic panels, the choice fell on the panels described in [25], which are highperformance polycrystalline units, connected to the load network through a DC/AC inverter (see Figure 1). On the other hand, in order to have a lower cut-in wind speed, vertical axis technology was adopted for wind turbines [26]. Table 1 provides the main technical specifications for the assumed PV and WT modules, which are referred to when running the optimal sizing procedure, as discussed later on in Section 3.2). Differently from the RES, for the rSOC system the energy conversion performance, in both SOFC and SOEC operation, that correspond to maximum power operation were assumed by referring to the relevant contribution found in the literature [27].
For the electrical storage system, the most widely adopted performance index, i.e., rSOC roundtrip efficiency (RTE), is expressed as follows [28]: The illustrated trajectory was initially extracted from 24 and then adapted to suitably represent the average daily profile of dwelling electricity demand [24].
As for the photovoltaic panels, the choice fell on the panels described in [25], which are high-performance polycrystalline units, connected to the load network through a DC/AC inverter (see Figure 1). On the other hand, in order to have a lower cut-in wind speed, vertical axis technology was adopted for wind turbines [26]. Table 1 provides the main technical specifications for the assumed PV and WT modules, which are referred to when running the optimal sizing procedure, as discussed later on in Section 3.2). Differently from the RES, for the rSOC system the energy conversion performance, in both SOFC and SOEC operation, that correspond to maximum power operation were assumed by referring to the relevant contribution found in the literature [27].
For the electrical storage system, the most widely adopted performance index, i.e., rSOC round-trip efficiency (RTE), is expressed as follows [28]: where V SOFC/SOEC and I SOFC/SOEC are the fuel cell voltage and current, respectively, and ∆t SOFC/SOEC is the overall operating time in SOFC/SOEC mode. P BOP terms in Equation (1) are representative of the typical ancillary power absorptions (mainly due to the air blower) of a generic rSOC system [29].

Methodological Approach
The modeling and optimization tools, as developed and deployed in the proposed preliminary design of an rSOC-based renewable microgrid, are synthetically described and motivated hereinafter. It is worth remarking here how the main objective in designing the HST is to find the best compromise between costs and energy resiliency.

Models Description
In the following subsections, the PV and WT yield estimation methods and the simplified sizing-oriented modeling of an rSOC are proposed, mainly aimed at guaranteeing the best compromise between precision and computational time when executing constrained-optimization algorithms.

PV Yield Estimation
Insolation and wind profiles of a given location were retrieved from the public domain. The optimization analysis was particularly focused on the simulation of a specific case study in the location of Policastro Bussentino (SA) (see Table 1), characterized by relatively high average contributions of insolation and wind. As for the former yield, the following formula was adopted [22]: It is worth remarking how Equation (2) is valid for horizontal panels. The unavoidable losses [30] due to bad weather conditions (e.g., rainy and cloudy days) are here accounted for through the γ factor shown in the first Equation (2), which is evaluated on a monthly basis by comparing model outputs with experimental values acquired for the selected location from [31].

Modeling of Wind Turbine Power
The annual wind speed trajectory was found for 2011 in [32]. Then, the corresponding power generated by WT is estimated via Equation (3), whose details can be retrieved from the literature [33], and referring to Table 1 data. Once again, it is worth remarking how the modeling approach was adopted coherently with the objective of developing an optimization tool aimed at preliminarily designing an rSOC-based microgrid capable of guaranteeing appreciable self-sufficiency characteristics, which in turn would allow for the meeting of energy resiliency requirements e.g., in remote areas.

Literature-Derived Model of rSOC Energy Conversion Performance
Equations (4) and (5) are used to evaluate, as a function of the efficiency in the current operating mode, the rSOC fuel production/consumption: .
It is worth noting that a conservative estimation of effective RTE rSOC is performed, that is to say that all the SOEC heat demand [29] is met by burning hydrogen. Assuming 0.95 burner efficiency, the above operating strategy causes the original η SOEC value, i.e., 0.76 in 28, to reduce down to 0.64 (see Equation (4)).
The above introduced rSOC efficiency values [27] correspond to the nominal operating point (i.e., 28 kW of electric power) of a high-pressure operating rSOC system. In this work, it is assumed that such values can be safely extended to different rSOC sizes of the same technology, provided that:

•
The rSOC is operated at constant power in both SOEC and SOFC mode; • The assumed efficiency data are scalable up or down to the size determined by the optimization task, as described in see Section 3.2.
In order to cope with the first assumption above, more detailed sizing analyses shall be performed, particularly aimed at assessing the opportunity of integrating the proposed rSOC unit with appropriate hybridizing devices, such as super-capacitors and batteries, to suitably manage highly fluctuating power requests [22]. This aspect is of particular relevance as it is well known how abrupt and frequent changes in operating condition may be very detrimental to SOC units as a consequence of the induced dangerous thermal-stresses [34].
The daily variation of HST state of charge (SOC HST , intended as the ratio between actual and maximum hydrogen mass storage m H2,HST ) is estimated as follows:

Economic Data and Analysis
In this study, the simple payback (PB) was selected to perform techno-economic feasibility assessment of the rSOC-based renewable microgrid: where: (i) TC is total microgrid cost, evaluated by including all main component costs (i.e., PV, WT, rSOC, HST), (ii) IC is the installation cost, here assumed equal to 10% of the TC; (iii) MC is the annual maintenance cost (i.e., under the hypothesis it equals 3% of TC; (iv) TD is the tax deduction, assumed equal to 50% of TC, according to Italian governmental rules [35]) and (v) ABC is the annual bill cost, which should be paid by involved owners without installing the proposed RES + rSOC microgrid.
Since the plant is grid-independent, the latter variable is considered a revenue, here estimated assuming that the average cost of energy (in Italy) is 0.25 €/kWh. For the analysis of the photovoltaic panels and wind turbines costs, specific manufacturer data-sheets were considered (see corresponding data in Table 2). Finally, the cost of reversible solid oxide fuel cells was set to 340 €/kW [36], while the cost of storing the hydrogen gas was assumed equal to 400 €/kg [37].

Constrained Optimization Algorithm
The microgrid sizing was carried out via a constrained optimization task, here expressed by Equation (8). The optimization task was implemented and run in the Excel ® spreadsheet environment [38]. Particularly, the default resolution method (i.e., the generalized reduced gradient (GRG) [39]) was adopted. Beyond minimizing the simple pay-back time, another key objective was to suitably account for the impact of renewables and rSOC interaction in terms of SOC HST daily variation. As for the latter aspect, the goal was to guarantee, through the constraint c 4 defined later on in this section, that final (i.e., end day of the year) and initial (i.e., first day of the year) tank filling levels do not differ, thus ensuring year-through charge sustaining management of HST. The variables to be optimized are: nominal PV power (P PV ), nominal wind turbines power (P W ), and m H2,HST : The optimization is subject to the following ci constraints: where c 4 ensures that renewable energy never gets wasted. It is also worth remarking here that day i and day f in the c 5 relationship relate to January the 1st and December 31st, respectively, whereas c 2 and c 3 shall hold valid for the entire year, thus limiting both hydrogen shortage and excess. This will in turn avoid, respectively, running out of fuel in the case of sudden grid unavailability, and wasting renewable energy in the case of unpredicted surplus. As for constraint c 4 , it is worth anticipating here that the baseline optimization outcome (i.e., the one meeting the c 3 constraint, as detailed in the section below) ensures fully avoiding any waste of renewables, since the ratio between yearly maximum renewable power surplus and SOEC nominal power is safely limited below 90% (i.e., 89.7%).

Presentation and Discussion of Optimization Results
In this section, the outcomes of the constrained optimization task (see Section 3.2) are presented and discussed. Table 3 shows the sizes of the main microgrid components and the minimum PB obtained through the Excel solver (with default settings). Figure 3, which shows the simulated SOC HST trajectory, demonstrates the reliability of the HST charge sustaining management resulting from the constrained optimization procedure.  . Simulated year-through solid oxide cell (SOC)HST trajectory. A and B labels highlight minimum and final hydrogen tank levels, respectively, which are used later on in the paper to assess the potential self-sufficiency range achieved by the optimal renewable rSOC-based microgrid design. According to Table 3, the values of nominal powers and of the hydrogen storage tank size determine the plant cost, while the other costs are obtained from the hypotheses listed in Section 3.1.4. On the other hand, Equation (7) allows obtaining the payback value to be included in the cost function (see Equation (8)). The optimal value yielded on output by the constrained-optimization task interestingly sets to 10.7 years, which increases by 15% if rSOC capital expenditures (CAPEX) costs increase up to 680 € per kW. Figures 4 and 5 show the simulated photovoltaic and wind power supply trends, indicating the physical coherence of the assumed simplifying hypotheses and modeling approach, accounting for both daytime and seasonal effects. It particularly demonstrates how the selected location is well served by both insolation and wind energy sources, as the lack of the former is fully compensated by the latter availability and vice versa, depending on the time of day. As for the effective contribution of renewable sources, Figure 5 highlights how WT power is often much lower than its corresponding nominal value (see Table 3), whereas PV panels appear better exploited from this point of view, as shown in Figure 4. In order to deepen such an aspect, the distribution of WT power contribution throughout the year shall be analyzed. Particularly, the cumulative curve associated with such a distribution, shown in Figure 6, along with yearly wind speed profile, indicates that WTs often work in off-design conditions. The latter aspect can surely be of high importance in terms of plant cost impact. It is indeed expected that grid-connected microgrids shall be preferred in such locations. Nevertheless, the objective of providing a methodology capable of yielding on output a potentially fully grid-independent microgrid configuration was advantageous. Of course, the opportunity to rely on more favorable locations (in terms of wind speed availability and exploitability) is expected to play an important role in obtaining more cost-effective designs. . Simulated year-through solid oxide cell (SOC) HST trajectory. A and B labels highlight minimum and final hydrogen tank levels, respectively, which are used later on in the paper to assess the potential self-sufficiency range achieved by the optimal renewable rSOC-based microgrid design. Table 3, the values of nominal powers and of the hydrogen storage tank size determine the plant cost, while the other costs are obtained from the hypotheses listed in Section 3.1.4. On the other hand, Equation (7) allows obtaining the payback value to be included in the cost function (see Equation (8)). The optimal value yielded on output by the constrained-optimization task interestingly sets to 10.7 years, which increases by 15% if rSOC capital expenditures (CAPEX) costs increase up to 680 € per kW. Figures 4 and 5 show the simulated photovoltaic and wind power supply trends, indicating the physical coherence of the assumed simplifying hypotheses and modeling approach, accounting for both daytime and seasonal effects. It particularly demonstrates how the selected location is well served by both insolation and wind energy sources, as the lack of the former is fully compensated by the latter availability and vice versa, depending on the time of day. As for the effective contribution of renewable sources, Figure 5 highlights how WT power is often much lower than its corresponding nominal value (see Table 3), whereas PV panels appear better exploited from this point of view, as shown in Figure 4. In order to deepen such an aspect, the distribution of WT power contribution throughout the year shall be analyzed. Particularly, the cumulative curve associated with such a distribution, shown in Figure 6, along with yearly wind speed profile, indicates that WTs often work in off-design conditions. The latter aspect can surely be of high importance in terms of plant cost impact. It is indeed expected that grid-connected microgrids shall be preferred in such locations. Nevertheless, the objective of providing a methodology capable of yielding on output a potentially fully grid-independent microgrid configuration was advantageous. Of course, the opportunity to rely on more favorable locations (in terms of wind speed availability and exploitability) is expected to play an important role in obtaining more cost-effective designs.   Table 1); (b) Histogram of the yearly wind speed profile.

According to
In the event that neither grid electricity nor renewable energy are available, it is possible to calculate the resiliency that the microgrid rSOC guarantees in terms of self-sufficiency, considering electricity demand (see Figure 2), the hydrogen consumed in SOFC mode (see Equation (5)), and the storable chemical energy in the HST (see Table 3). Assuming that the load profile is constant and    Table 1); (b) Histogram of the yearly wind speed profile.
In the event that neither grid electricity nor renewable energy are available, it is possible to calculate the resiliency that the microgrid rSOC guarantees in terms of self-sufficiency, considering electricity demand (see Figure 2), the hydrogen consumed in SOFC mode (see Equation (5)), and the storable chemical energy in the HST (see Table 3). Assuming that the load profile is constant and    Table 1); (b) Histogram of the yearly wind speed profile.
In the event that neither grid electricity nor renewable energy are available, it is possible to calculate the resiliency that the microgrid rSOC guarantees in terms of self-sufficiency, considering electricity demand (see Figure 2), the hydrogen consumed in SOFC mode (see Equation (5)), and the storable chemical energy in the HST (see Table 3). Assuming that the load profile is constant and  Table 1); (b) Histogram of the yearly wind speed profile.
In the event that neither grid electricity nor renewable energy are available, it is possible to calculate the resiliency that the microgrid rSOC guarantees in terms of self-sufficiency, considering electricity demand (see Figure 2), the hydrogen consumed in SOFC mode (see Equation (5)), and the storable chemical energy in the HST (see Table 3). Assuming that the load profile is constant and equal to that already considered previously, it is possible to estimate the self-sufficiency of the plant at time t in the case of lack of renewable electricity production, as follows: The previous equation allows the assessment of whether the HST is capable of fully meeting user request for a time interval ranging from almost 1.62 days, i.e., in correspondence with the minimum HST state of charge labeled as point A in Figure 3, up to 5.4 days, this latter being associated with a half full HST (i.e., point B). Such an outcome confirms the highly promising resilient features the proposed renewable rSOC-based microgrid can guarantee.
In order to further investigate the impact of design criteria and assumptions on constrained optimization outcomes and, in turn, on the real world implications of different configuration scenarios, a parametric analysis was carried-out by varying the constraint (see Figure 7) on minimum SOC HST (i.e., SOC HST,min from now on) between 5% and 25%. These effects were assessed by running two further optimization analyses, whose outcomes are analyzed and discussed against the previously presented baseline (i.e., the one corresponding to SOC HST,min ≥ 15%). Figures 7 and 8 display these results and provide interesting indications. Relaxing the constraint on minimum hydrogen storage (e.g., setting SOC HST,min ≥ 5%) causes plant cost, simple payback, HST mass, and self-sufficiency range to reduce, as shown in Figure 7. Particularly, self-sufficiency exhibits (see Figure 8) a nonlinear relationship with SOC HST,min ; low SOC HST,min , obtained as mentioned before by relaxing the constraint c 3 in Section 3.2, causes the difference between the minimum and maximum range of self-sufficiency to increase. On the other hand, strengthening such constraints would determine the opposite behavior, thus allowing larger self-sufficiency independently from when the emergency event (e.g., simultaneous unavailability of RES and electricity grid) occurs. Nevertheless, considering the higher capital costs and paybacks associated with an over-strengthened c 3 constraint, SOC HST,min = 15% can be selected as a suitable design criterion to achieve the best compromise between techno-economic feasibility and resiliency requirements for the investigated rSOC microgrid. It is also worth pointing out that SOC HST is never higher than 60%, as shown in Figure 3, mainly because the designed rSOC microgrid does not necessarily interface with the electricity grid, thus making it more difficult to cope with SOC HST,min requirements in case a small tank size is selected. Therefore, the selected design criteria, particularly those related to microgrid resiliency and self-sufficiency range, shall be revised or partially relaxed if a more effective exploitation of the HST has to be achieved, which in turn could also play a positive role in reducing overall plant CAPEX. To further support the above comment, an additional optimization analysis was performed by assuming initial HST to be equal to 0.7. It was verified that in this way the PB could be slightly reduced (5%) and an almost full exploitation of the selected HST level range can be achieved, while causing, as the main drawback, maximum self-sufficiency to reduce down to four days, whereas current design criteria allows achieving up to six and a half days (see Figure 8).
As for rSOC operating conditions, it is important to remember that heat is produced in SOFC mode, thus leaving further design margins. For instance, a thermal storage unit could be included in the overall rSOC BoP, in such a way as to enable re-using the heat stored in SOFC mode to supply the necessary heat in SOEC phases. Nevertheless, such a component addition may have a negative impact on rSOC BoP complexity and would require careful attention to heat exchanger design and development, which in turn could result in higher CAPEX and complex energy management. Finally, the application under analysis could also be used to supply a hydrogen car refueling station. Considering the Toyota Mirai specifications [40] and particularly the fuel economy (as high as 100 km/kg) and tank capacity (5 kg), the designed HST (see Table 3), if assumed half-full, could provide up to seven cars with the amount of fuel required to drive a distance of 500 km.

Conclusions
The main target of the presented research was the development of models and procedures aimed at preliminarily designing and evaluating the technical feasibility of an rSOC-based renewable microgrid, for both residential power supply (even in islanded configuration) and hydrogen supply to industrial processes and refueling stations. Reaching these goals can be reached by the main innovation introduced, namely the achievement of charge sustaining energy management for the hydrogen storage tank. This strategy ensures that the tank will neither get filled-up nor ever become empty during the year, thus allowing the full exploitation of the available RES plant. It was estimated that the investement could be fully recovered in about ten years, while the resulting optimal RES plant sizes appear reasonable with respect to targeted average load demand. Beyond confirming the high techno-economic potential of an adequate combination of RES and rSOC, optimization outcomes also demonstrate the possibility, by adopting the proposed procedure, of simultaneously addressing the interconnected needs of optimal system design and rSOC management.
The specific analysis, carried-out on the impact of strengthening the constraint on minimum hydrogen storage (i.e., by imposing higher minimum admissible hydrogen levels in the tank), indicated that larger self-sufficiency (i.e., the capacity of meeting load demand independently from RES and grid availability) can be attained, while admitting an acceptable CAPEX increase. Moreover, larger self-sufficiency, which in turn determines larger hydrogen storage tanks, introduces further deployment opportunities, such as the coupling of residential loads to refueling (i.e., for hydrogen-

Conclusions
The main target of the presented research was the development of models and procedures aimed at preliminarily designing and evaluating the technical feasibility of an rSOC-based renewable microgrid, for both residential power supply (even in islanded configuration) and hydrogen supply to industrial processes and refueling stations. Reaching these goals can be reached by the main innovation introduced, namely the achievement of charge sustaining energy management for the hydrogen storage tank. This strategy ensures that the tank will neither get filled-up nor ever become empty during the year, thus allowing the full exploitation of the available RES plant. It was estimated that the investement could be fully recovered in about ten years, while the resulting optimal RES plant sizes appear reasonable with respect to targeted average load demand. Beyond confirming the high techno-economic potential of an adequate combination of RES and rSOC, optimization outcomes also demonstrate the possibility, by adopting the proposed procedure, of simultaneously addressing the interconnected needs of optimal system design and rSOC management.
The specific analysis, carried-out on the impact of strengthening the constraint on minimum hydrogen storage (i.e., by imposing higher minimum admissible hydrogen levels in the tank), indicated that larger self-sufficiency (i.e., the capacity of meeting load demand independently from RES and grid availability) can be attained, while admitting an acceptable CAPEX increase. Moreover, larger self-sufficiency, which in turn determines larger hydrogen storage tanks, introduces further deployment opportunities, such as the coupling of residential loads to refueling (i.e., for hydrogen-

Conclusions
The main target of the presented research was the development of models and procedures aimed at preliminarily designing and evaluating the technical feasibility of an rSOC-based renewable microgrid, for both residential power supply (even in islanded configuration) and hydrogen supply to industrial processes and refueling stations. Reaching these goals can be reached by the main innovation introduced, namely the achievement of charge sustaining energy management for the hydrogen storage tank. This strategy ensures that the tank will neither get filled-up nor ever become empty during the year, thus allowing the full exploitation of the available RES plant. It was estimated that the investement could be fully recovered in about ten years, while the resulting optimal RES plant sizes appear reasonable with respect to targeted average load demand. Beyond confirming the high techno-economic potential of an adequate combination of RES and rSOC, optimization outcomes also demonstrate the possibility, by adopting the proposed procedure, of simultaneously addressing the interconnected needs of optimal system design and rSOC management.
The specific analysis, carried-out on the impact of strengthening the constraint on minimum hydrogen storage (i.e., by imposing higher minimum admissible hydrogen levels in the tank), indicated that larger self-sufficiency (i.e., the capacity of meeting load demand independently from RES and grid availability) can be attained, while admitting an acceptable CAPEX increase. Moreover, larger self-sufficiency, which in turn determines larger hydrogen storage tanks, introduces further deployment opportunities, such as the coupling of residential loads to refueling (i.e., for hydrogen-fed cars) and recharging station (i.e., for electric vehicles) demands. In this way, a highly potential synergy between clean stationary generation and sustainable mobility can be achieved, thus also contributing, through a leveraging effect, to speeding up the development of effective and properly functioning comprehensive and low carbon footprint smart energy systems. Research follow-ups, beyond including additional electrical and thermal loads to be met by the rSOC-based microgrid, will focus on refining the techno-economic feasibility assessment. Particularly, the total cost of ownership will be assessed, including the impact of interest rates, allowing a more accurate comparison between proposed renewable exploitation and other technology solutions. Furthermore, more detailed models will be used to perform model-based definition of on-field applicable control strategies, and to address the relevant aspects of rSOC thermal management and related additional degrees of freedom, in terms of additional loads (i.e., thermal) being met, as well as the opportunity to store and re-use (during the SOEC phase) the heat produced in SOFC mode. Lastly, both rSOC battery hybridization and interaction with the electricity grid will be explored to ensure acceptable self-sufficiency capabilities, while improving the exploitation of available hydrogen storage tanks.