Optimal technoeconomic reliability‐oriented design of islanded multicarrier microgrids with electrical and hydrogen energy storage systems considering emission concerns

Although several studies have been performed on energy systems, there is a gap in examining the impact of energy storage systems (ESSs) technologies on the technoeconomic design of optimal multicarrier microgrids (MCMGs), considering environmental concerns. This paper aims to address this research gap by developing an optimal technoeconomic reliability‐oriented design of MCMGs, specifically focusing on battery storage systems (BSSs) and hydrogen storage systems (HSSs). The study was applied to MCMG at an actual MCMG, using HOMER Pro software. Additionally, a reliability‐oriented sensitivity analysis is conducted to understand the effects of reliability constraints, such as desired maximum capacity shortage, on the performance of various ESS technologies. One of the main contributions is analyzing the HSS and BSS from different viewpoints, such as cost, emission, and other technical–economic features for MCMGs. The comparative test results show that if a highly reliable MCMG is desired, the BSS‐based MCMG is the more practical option in terms of technoeconomic indices. However, regardless of reliability constraints, the HSS‐based MCMGs offer better environmental conditions. The total net present cost of the HSS‐based MCMG is 1.68% higher than the BSS‐based one, while it has 17.99% less CO2 emissions, while the HSS one has 17.99% less CO2 emissions.


| INTRODUCTION 1.| Motivation and problem statement
Rapid population growth and urbanization are among the reasons for the increasing demand for energy. 1 This rise in energy demand raises concerns from technical, economic, and environmental perspectives. 2Expanding conventional power production and energy systems is not only costly but also has adverse effects on the environment. 3As a result, there has been a growing deployment of renewable energies to address the challenges posed by the increase in energy demand and global warming concerns. 4dditionally, new concepts and frameworks, such as multicarrier microgrids (MCMGs) and energy hubs, have been introduced to optimize solutions. 5In conventional microgrids (MGs), only electricity and electrical loads are typically considered.However, energy systems also include nonelectrical loads, such as heat loads, and can involve chemical energy carriers, like, natural gas.By managing electrical loads and heat loads within an integrated framework, additional benefits can be achieved.Optimal design, planning, and operation of both gas and electricity simultaneously lead to more effective solutions.In fact, simultaneous management of various energy carriers (electricity and gas), while considering the interconnections of electrical and heat demands, can result in efficient outcomes. 6he rapidly increasing energy demand leads to an inadequate power supply if the capacity of energy sources is not expanded.Additionally, increasing investment cost is a major barrier to expanding power production. 7herefore, optimal design and operation are crucial solutions to reduce the cost of energy (COE) systems. 8umerous studies have investigated the optimal planning and operation of MGs and MCMGs in both islanded and on-grid modes. 9The optimal sizing of standalone MCMGs has also been explored in various studies. 10urthermore, the optimal design of MGs and MCMGs, considering reactive power management, such as the selection and location of capacitor banks, as well as optimal reconfiguration, has been studied in Montoya et al. 11 While there has been significant attention given to optimizing nonconvex renewable sources, less focus has been placed on the optimal reliability-oriented design of MCMGs using various energy technologies.This research aims to address this gap by proposing an optimal reliability-oriented design of MCMGs based on different energy storage technologies.Additionally, this study examines the advantages of various energy storage technologies from a technoeconomic perspective, including the investigation of new storage technologies, like, hydrogen-based ones.

| Literature review
The optimal design of MGs and MCMGs has been extensively studied in previous research. 12One important issue in renewable-based MCMGs is the integration of energy storage systems (ESSs) to address the challenges posed by the intermittencies of renewable distributed generation. 13With the increasing penetration of renewable energies in energy systems, their dependency on weather and environmental conditions introduces uncertainty in power generation. 14This uncertainty affects both the optimal design and operation of MGs.Therefore, ESSs, such as battery storage systems (BSSs), are utilized to mitigate the negative impacts of these uncertainties. 15Savargaonkar et al. 16 reported a new uncorrelated sparse autoencoder with long shortterm memory to identify patterns relevant to the longterm state of charge's (SOC's) estimation due to complex battery cell characteristics, such as aging.In Lei et al., 17 a suitable look-ahead economic dispatch framework was presented for the wind-thermal-bundled power system based on the variable ramp rate of coal-fired units and the flexible load transfer strategy.Xie et al. 18 introduced a method for optimal capacity and type planning of the bundled wind-thermal generation system incorporating wind turbines (WTs).The optimal design of an energy system was studied in Saheb Koussa and Koussa. 19The system was modeled in HOMER software and included a WT, grid, and photovoltaic (PV).The goal was to evaluate the technoeconomic aspects of the proposed energy system considering greenhouse gas (GHG) emissions.Their study presented a comparison between the on-grid MG and a standard grid operation regarding the economic traits of both systems.The simulation results from two different locations showed a 30% and 35% GHG reduction and an 81% and 76% reduction in the COE in locations Salah and Adrar, respectively.The main gap in this study is that it did not include any storage technology.It could be argued that the presence of the grid makes any storage system obsolete; however, a properly sized storage system could participate in sellback to the main grid, which could benefit the cost of the overall system, the reliability of the grid, and peak shaving.Another research gap was related to the lack of thermal energy in the modeling, as it could be one of the main sources of GHG emissions.Furthermore, there was no discussion about the reliability of the system.In Ribó-Pérez et al., 20 an islanded MG, including a biomass gasfired generator, PV, grid, and battery was studied using HOMER software.The goal was to analyze the technoeconomic design of the proposed system for a clean power supply.The simulation was conducted in two case studies for two different remote locations, and the results of the energy cost were compared with the cost of grid extension.Their result indicated that the islanded hybrid MG was more economical than extending electricity grids to those areas.Similar to some of the previous references, this research did not include thermal energy power, reliability analysis, alternative storage systems, and the effects of emission on the optimal design and operation of MGs and their subsystems.
A study to find the optimal design and size configuration of an MG that consisted of PV, WT, diesel generator, and battery has been done by Singh et al. 21he goal was to find the optimal solution that satisfies the technical, social, and economic objective functions (OFs).The simulation and optimization were carried out using dynamic domain search, that is, the integration of particle swarm optimization (PSO) algorithm and differential evolution.The optimal design of the hybrid system was evaluated based on the OFs and designers' preferences.The results for the selected case study had a levelized COE of 0.799 $/kWh and a satisfaction level of 72.82% for all objectives.This research did not include thermal power supply, and there was no discussion of any alternative storage technology that could change the results of optimal size and emission levels.Furthermore, the effect of reliability constraints on the optimal system and problem objectives was not considered.
In other research in the area of optimal sizing, Bhuiyan et al. 22 studied a single objective optimization of an islanded MG to minimize the life cycle cost subjected to loss of power supply probability (LPSP).The case study MG included PV, WT, battery, and a diesel generator.The proposed algorithm in their study employed meteorological simulation and iterative techniques to find the optimal size of the components and power management strategies.The effect of emission penalties and thermal power demand were neglected in this study.Moreover, this study could benefit from the comparison between battery and hydrogen storage systems (HSSs).
Research that considered both thermal and electrical powers has been done by Kim et al. 23 The authors investigated the impact of combined cooling heat and power systems with and without PV for residential applications from the energy efficiency, economic, and environmental aspects.Then, the energy system was modeled in HOMER for different case studies of residential buildings in Atlanta.Their results included energy efficiency scenarios and economic analysis to improve the cost of the overall system.This research, however, did not consider the effect of emission penalties in their study, and since their system was assumed to be connected to the main utility grid, it also did not include any type of storage system.furthermore, the role of power supply reliability was neglected in their research.
In a similar study by Waqar et al., 24 a hybrid MG with combined heat and power (CHP), battery, and PV was studied in grid-connected and islanded modes in HOMER software as a solution to power outages in Pakistan.The objective of their study was to find the optimal configuration of the hybrid system to minimize the cost and maximize grid sales.The results of their study included the optimal design of the hybrid CHPbased system for proposed case studies.This research did not include any reliability indices for their simulation and optimization and also, the effects of using different storage systems were not studied.Finally, there was limited discussion on the emission penalties and their effects on the optimal design and operation of the CHP and MG.
BSS is one of the most commonly used types of ESSs and has been extensively studied in the literature to address the challenges posed by uncertainties and other technoeconomic issues in MGs and MCMGs.For example, the optimal sizing of an on-grid electrical MG consisting of PV, WT, and BSS was investigated in Akram et al. 25 The main objective of this study was to find the optimal design of the MG with minimum cost and maximum reliability.Another study presented the optimal sizing of a standalone MCMG that included a photovoltaic-thermal (PVT) system, WT, ESSs, and a boiler. 10Unlike other studies that focused solely on PVs, this study considered the use of PVT units, which combine PV technology with thermal energy capture.These units generate both electricity and heat simultaneously, but their overall efficiency tends to be lower compared with PV units due to the utilization of solar energy for heat generation. 10The evolutionary particle swarm optimization (E-PSO) algorithm has been used to minimize the total net present cost (TNPC) and COE.The E-PSO algorithm serves as inspiration for both evolutionary algorithms and the PSO algorithm.The concept behind E-PSO is to enhance the PSO scheme by incorporating an explicit selection procedure and selfadapting properties for its parameters.In a previous study, 26 a method was introduced to optimize the sizing of BSSs in a droop-controlled islanded MG.Amini et al. 27 presented a novel approach for modeling and optimizing the size of BSS in MGs.They optimized the planning and operation of MGs, including BSS, based on factors, such as BSS characteristics, lifetime, capacity degradation, and depth of discharge (DOD).The lifespan of a BSS refers to the total period it remains operational, while capacity degradation is directly linked to the gradual decrease in its ability to store and discharge energy.The degradation of capacity in a BSS is influenced by the number of charge and discharge cycles and the DOD.These issues were thoroughly examined in their study. 27Zolfaghari et al. 28 investigated the reduction of operational costs in MGs through the optimal sizing of BSS using convex optimization methods.In another study, 29 an MG consisting of renewable and nonrenewable energy sources, along with a BSS, was analyzed to determine the optimal size of its components for a distribution network in the Amazon.The primary objective of this study was to minimize the TNPC and COE.The proposed nanogrid provided reliable power with high penetration of renewable energy for residential sectors.Zolfaghari et al. 28 utilized the Grasshopper Optimization Algorithm to find the optimal design of a BSS-based networked MG that incorporated PVs, WTs, and diesel generators.The main goal of their study 30 was to minimize the COE while considering the probability of power supply deficiency.The application of the studied MGs was in five residential units in Nigeria.
In most of the references discussed, the focus has been on BSS, without considering the impact of other energy technologies.Additionally, the interactions between reliability, environmental, and technical indices have not been studied in this category of research.A storage technology that has gained recent attention is HSSs. 31HSS typically consists of an electrolyzer, fuel cell, and hydrogen tank with various configurations.Similar to BSS, HSS has seen increased application with the development of hydrogen vehicles (HVs) in addition to electric vehicles (EVs). 32Hydrogen also has various industrial applications, such as oil refining, ammonia production, methanol production, and steel production.This wide range of applications creates opportunities for sites deploying HSS to sell hydrogen to nearby industries.
It is known that the responsiveness of HSS is less than that of BSS.However, because of hydrogen's considerable energy density, lower cost on larger scales, longer life cycle, and harmless by-products of such systems, HSSs have been widely studied in many literatures. 33For example, the optimal planning of an HSS was researched for cross-regional consumption of renewable energy considering the uncertainty. 34Qiu et al. 34 included two layers of the planning model to determine investment and operation costs.The first layer of their model determines the feasible planning approach, while the second layer determines the optimal operation of the system based on robust optimization and uncertainty.In a similar study to the proposed study in this article, a mixed integer linear model of an HSS was formulated to find the optimal planning of a hydrogen-based multienergy system. 35The purpose of this was to minimize the annual capital cost and operation cost of the system under different demand profiles.In another study, a robust energy management tool was developed for islanded MGs. 36The presented system included an HSS and considered demand response.Their findings showed that having a flexible demand is more economical than employing an HSS.In conclusion, although the HSS might not have responsiveness as good as BSS, its advantages derived researchers to investigate many systems, from vehicles to MGs, to find better solutions for energy cost, emission, and reliability.Moreover, the rapid advancements in HSS technology may someday remove or reduce the effects of this barrier.
The optimal design of MGs and MCMGs using HSS was studied in existing research.Hadidian Moghaddam et al. 37 investigated the optimal sizing of a standalone hybrid PV/WT system with HSS using the flower pollination algorithm.The optimization considered loss of energy expected and load of loss expected indices, which measure power shortage and load loss in the system.Zhang et al. 38 examined the optimal sizing of a hybrid off-grid MG with HSS using chaotic search, harmony search, and simulated annealing algorithms.Jamshidi and Askarzadeh 39 conducted an optimization on different configurations of hybrid MGs, consisting of both BSS and HSS, to minimize the COE.The comparison showed that implementing both BSS and HSS in an MG is advantageous compared with HSS-only systems for long-term investment.A multiobjective optimization of hybrid energy systems, including PV, diesel generator, and HSS, was conducted to find the optimal design for islanded applications. 39The impact of variables such as emission cost and component cost on the optimal sizing was evaluated.The results showed that HSS reduced the TNPC.Wen and Aziz 1 tried to solve the uneven temporal-spatial distribution of renewable energy sources, deploying green hydrogen and ammonia as short-and long-term energy storage mediums.The synergistic impact of these two storage media was evaluated in an integrated renewable multigeneration system.Wang et al. 40 presented a new multiobjective capacity programming and operation optimization for energy systems.In the study of the integrated energy system by Wang et al., 40 HSS was considered for collective energy communities.In Wang et al., 41 the operation of integrated energy systems was optimized, using the HSS and cooperative game approach.The datadriven energy management system was reported in Wen and Aziz 42 for the flexible operation of energy hubs and MCMGs, while hydrogen ammonia technologies were studied.Abdelghany et al. 43,44 reported a new strategy according to the hierarchical rolling horizon control and model predictive control (MPC) to efficiently manage an HSS-based MG.Abdelghany et al. 45 reported the control platform architecture of a real HSS-based energy system paired with WTs located in north Norway.
The optimal planning and operation of an MCMG were presented in Nosratabadi et al., 46 which included various storage systems.The findings showed significant energy savings in both winter and summer seasons.However, similar to the first category of reviewed papers on BSS, the impacts of different energy technologies compared with HSS have not been examined in the second category of reviewed papers on HSS.Additionally, the interactions between reliability, environmental, and technical indices have not been studied for MGs or MCMGs equipped with HSS.
Table 1 provides a summary of the literature review conducted on the optimal design of MGs and MCMGs.Numerous studies have been conducted on this topic;

| Main contributions and structure of the paper
As discussed in the literature review, it should be noted that different studies worked on the optimal planning of MGs.Some available research works studied the electrical demand of a sample community but did not include the thermal demands of such applications.Some studies considered only the battery or the HSSs but did not compare the technoeconomic aspects of these two systems.Many studies investigated the optimal operation of CHP systems but only a few have studied the optimal size and operation of the CHP systems simultaneously.Some references have the reliability of power supply as a constraint, and some do not; however, only a few have considered the effects of different reliability constraints on MGs' optimal operation and planning.From the environmental point of view, some studies have considered the emission penalties but there was a research gap in comparing the effects of emission penalties on fuel consumption of the MG, CO 2 reduction of the MG, optimal size and operation of the system, the trade-off between the costs of the overall system (MG costs and emission penalties), and the effects of emission penalties on these factors respect to different storage system technologies.The other part of the available studies is related to renewable power generation, mainly wind and solar.However, not too many references studied the optimal system when these renewable sources are coupled with alternative fossil fuel-based systems (CHP) from the operation and planning points of view.
This study investigates all these research gaps in a multiscenario-based simulation to provide a clear and comprehensive insight into optimal MCMG design and operation concerning the effects of storage technology, system reliability, and emission policies.To the best of our knowledge, there is no other study that thoroughly investigated different power generation systems in a single optimization problem to achieve an overall optimal answer for the presence of emission taxes in various reliability constraints.Furthermore, the complete comparison between two of the most recognized storage systems provides valuable information on technical and economic sections of sustainable development that have not been done before.
This research aims to address the existing research gaps by proposing an optimal design approach for islanded MCMGs that considers both technoeconomic and reliability aspects.The study focuses on the integration of different ESS technologies, such as BSSs and HSSs.Briefly, the main contributions, advantages, and technical features of this research can be summarized as follows: One significant difference between this research and existing studies is the examination of a university campus as an MCMG, which has received less attention in previous research works.The HOMER software is utilized for implementing this research.
The remaining sections of this article are structured as follows: Section 2 presents the MCMG structure under study, incorporating hydrogen ESS and electrical ESS.Section 3 introduces the modeling approach for the proposed reliability-oriented design of MCMGs, considering both electrical and hydrogen ESSs.Additionally, Section 4 provides details about the test system and simulation results.Finally, Section 5 concludes the article.

| STRUCTURE OF THE MCMG UNDER STUDY
Figure 1 illustrates the structure of the MCMG that is examined in this study.The MCMG incorporates both HSS and BSS.The main focus of this research is to investigate the optimal design and planning of MCMGs in arid climate conditions.However, it should be noted that the proposed research can also be applied to other climate conditions.The case study for this research is based on an actual test system located at the University of Kashan in Kashan, Iran.The simulation considers the potential emission penalties associated with CO 2 , as well as excess NO and SO 2 emissions.Additionally, a sensitivity analysis is conducted to assess the impact of critical parameters on the yearly maximum capacity shortage (MCS) in the reliability-oriented design and planning of MCMGs.The MCS, which stands for the MCS, is a reliability index expressed as a percentage.It is calculated by dividing the unmet electrical load by the total energy demand per year, measured in kilowatthours (kWh).This index can be used to optimize the design of MGs or MCMGs.
It should be noted that the reliability assessment of an energy system network is an essential part of planning research, and there is much literature available that assesses the reliability of various systems.The HOMER Pro cannot provide a detailed reliability assessment of the network.However, studying the reliability of the proposed hybrid energy system in detail and completely specified is not the main scope of this article.Nevertheless, it has been tried to address the reliability indices, using the MCS index, according to the capabilities of the HOMER Pro.The proposed MCMG in this study will be simulated in different MCS levels for each of the storage technologies and emission scenarios.Also, the results of all reliability constraints (different MCS levels) are compared in the case of cost, emission, excess heat generation, load loss, and so forth.However, there is room to specifically investigate the reliability of the proposed system and scenarios in detail in future studies.
In the case of the MCMG studied here, the power generation components include PV panels, WTs, and a CHP system for electrical power generation.The CHP system also includes a boiler for thermal power generation.Excess electrical power generated by the CHP, PV, or WT is stored in the battery, which is part of the electrical ESS considered in this study.Alternatively, excess electrical power is used by the electrolyzer to generate hydrogen, which is then stored in the hydrogen tank as part of the hydrogen ESS (HSS).During peak periods, the battery and hydrogen tank discharge their stored energy to supply sufficient power to consumers.If hydrogen-based ESS is used, the fuel cell component consumes stored hydrogen from the tank to generate power.On the basis of the output power of renewable sources, the CHP system supplies the remaining electrical demand using optimal sizing and operation of each component, considering emission taxes, gas prices, and operational costs.The heat recovery system of the CHP provides the necessary thermal demand.If the thermal demand exceeds the recovered heat from the CHP or if there is only thermal demand and it is not cost-effective for the CHP to operate at a specific time step, the boiler is responsible for supplying the remaining thermal power to ensure 100% thermal reliability.
In this paper, a standalone MCMG is studied.This study has focused to reach the reliability levels in a standalone mode and meeting the heat and electrical demands.In this study, the impacts of ESS and its different technologies (BSS and HSS) are highlighted.It should be noted that in future works, the feasibility of bidirectional power flow in the MCMG could be evaluated.In the grid-connected MCMGs (if it is feasible and there are suitable connection points between the MCMG and the upstream grid), the grid supports the reliability requirements.On the other hand, the economic advantages of the ESS could be examined.

| METHODOLOGY AND MATHEMATICAL MODELING
The mathematical modeling and characteristics of each component of the studied MG are presented in this section.

• PV
The output power of PV is calculated using (1), 55 where Y PV and f PV represent the rated capacity of the PV array and PV's derating factor, respectively.G T and T c are the solar irradiation and PV cell's temperature, while G T STC , and T c , STC represent the same variables at standard test conditions (STCs).
By neglecting the effects of temperature on PV's output power, the output power of the PV system can be simplified, as shown in (2).The PV manufacturers rate the output power of the modules at STCs.In STC, a radiation of 1 kW/m 2 , a cell temperature of 25°C, and no wind are assumed to calculate the output power of the PV modules.
• WT The WT output power is calculated using (3)-( 5). 56irst, the wind speed at the turbine's hub height is calculated using (3).The U hub and U anem variables represent the wind speed at the WT's hub height and anemometer height, respectively, while z hub and z anem are the WT's hub height and anemometer's height.
The output power of WT is calculated according to its power curve.Equation ( 4) represents the modeling for the output power of the WTs. 57In Equation ( 4), the wind speed at the WT's hub height should be used to increase the accuracy of the calculations.U, U ci , and U co represent the rated, cut-in, and cut-out wind speeds in (4).Also, P WT STP , represents the output power of the WT in standard temperature and pressure conditions (0°and 1 atm of pressure).
The WT generates no power when the wind speed is higher than the cut-out or lower than the cut-in wind speeds.On the other hand, the WT generates its rated power whenever the wind speed is between the rated and cut-out values.Otherwise, the output power of the WT is a function of the wind speed (a portion of its rated capacity).In addition to the WT's height and wind speed, the other factor affecting WT's power output is the air density, which is considered by (5), where P WT (U) and P WT,STP (U) are the WT's output power at actual and standard temperature and conditions, respectively, and ρ and ρ 0 represent the actual and standard air density.The standard air density is 1.225 kg/m 3 , and it is recorded at 15°C.
• Electrical ESS The Li-ion battery was chosen for this study due to its versatility, cost-effectiveness, and long lifespan.The power output modeling is presented in Equations ( 6)-( 12), 58 and the mathematical modeling of the battery and BSS is based on its maximum/ minimum charging/discharging power and SOC.The calculation of the maximum charging power of the battery and BSS should be carried out according to Equations ( 6)- (8).In this article, the kinetic battery model (KBM) is utilized to model the BSS.This model is used to determine the energy that can be stored or discharged from the ESS within a given time interval.The KBM is a two-tank model with kinetics.The first tank contains "available energy," which can be readily transformed into electrical energy, while the second tank holds "bound energy," which is chemically bound and not immediately accessible.
Three key parameters are used to illustrate the functioning of this two-tank model.The maximum/ theoretical storage capacity (Q max ) represents the total amount of energy that the two tanks can contain.The capacity ratio (c) is defined based on the capacity of the first tank to the total capacity of both tanks.The rate constant (k) is introduced to represent the conductance between the two tanks, while a third-rate constant shows how quickly bound energy can be converted to available energy.
The KBM is presented using (6) for the maximum charging power of the battery and BSS.In (6), Q 1 and Q represent the available and total energy at the beginning of each time step (Δt), respectively.The two-tank model for the BSS could be useful to understand how (6) works.The discussed three parameters of the KBM for the BSS have been used in (6).
Moreover, the maximum charging power of the BSS is limited by (7), where Q max is the total energy of the battery, and α c is the maximum charge rate (A/Ah).
( ) Furthermore, the maximum charging power of BSS is a function of the maximum current according to (8).In (8), N batt , I max , and V nom represent the total number of batteries in the BSS, maximum allowed charging current, and nominal voltage, respectively.
Finally, the maximum charging power of the BSS will be the minimum values of ( 6)-( 8), as shown in (9), while all constraints have been met.In (9), η batt,c is the charging efficiency and is calculated according to (10), where η batt,rt is the battery round trip efficiency, and η batt,c represents the charging power efficiency of the BSS.
Similarly, the maximum discharge power of the BSS is calculated according to (11) and (12).Similar to (6), the maximum discharge power of the battery should be calculated based on the KBM, as described in (11). 59The discharge power efficiency (η batt,d ) is modeled according to (12).
To clarify the functionality of the ESS model, it is important to note that the maximum absorption capacity of the ESS is determined for each time interval.This maximum charge power is utilized when there is excess power due to renewable energy sources or other factors.The maximum power absorption of the ESS is dependent on its SOC and its charge/discharge history.Three constraints and technical limits are considered to determine the maximum charge power for the ESS.
The KBM, represented by Equation ( 6), addresses the first constraint, which is the limitation imposed by the two-tank model on the maximum power absorption of the ESS.The second technical constraint is related to the maximum charge rate of the ESS, and this constraint is directly linked to the unfilled capacity, as demonstrated in Equation (7).Additionally, the maximum charge current serves as another technical limit for the ESS, as expressed in Equation (8).It is crucial for all these technical limits and constraints to be satisfied to determine the maximum storage charge power.Equation ( 9) should be applied to ascertain the amount of storage charge power that adheres to all technical limits.

• CHP and Boiler
The CHP in this study is assumed to operate with a reciprocating engine, and the electrical and thermal output power of CHP, as well as the thermal power output of the boiler, are modeled according to ( 13)-( 15), 5,53 where P E CHP and P T CHP are the CHP's electrical and thermal power output based on the natural gas input (M NG CHP ) and the electrical and thermal power efficiencies.The unit for natural gas input is m 3 per simulation step size (1 h).In ( 13)-( 15), P T B is the thermal power output of the boiler and M D B represents the boiler's diesel fuel intake.
• Hydrogen storage system The HSS comprises an electrolyzer, a fuel cell, and a hydrogen tank.The excess electricity in the BSS-based MCMG will be stored directly in the battery.However, in the MCMG equipped with HSS, the electrolyzer first consumes the excess electricity to produce hydrogen.Then, the hydrogen is stored in the hydrogen tank, and the fuel cell will use the stored hydrogen to generate adequate power if it is needed.The mathematical modeling of electrolyzer, fuel cell, and hydrogen tank are presented in ( 16) and (17). 38) ) As shown in ( 16) and ( 17), the input power of the electrolyzer is the summation of the renewable energy sources and the electrical power of CHP, multiplied by the electrolyzer's efficiency.Indeed, η EL , η I , and η V are the efficiency of the electrolyzer, the current efficiency, and the voltage efficiency, respectively.P EL-HT , P RES-EL , and P E EL CHP represent the hydrogen tank's input power from the electrolyzer (or the electrolyzer's total output), the electrolyzer's input power from the renewable sources, and the electrolyze's input power from the CHP, respectively.Converse to the electrolyzer, the fuel cell consumes hydrogen to generate power, and its | 2711 mathematical modeling is presented in (18). 60In (18), η FC represents the efficiency of the fuel cell.Also, P FC-INV and P HT-FC are the inverter input power from the fuel cell (fuel cell's total output) and the fuel cell's input power from the hydrogen tank, respectively.
The hydrogen tank modeling is based on the amount of hydrogen stored in the hydrogen tank, the amount of hydrogen consumption by fuel cell, and the amount of hydrogen generation by the electrolyzer.However, since the mathematical modeling of the electrolyzer and fuel cell are presented based on their input and output power, the mathematical modeling of hydrogen tank is also presented based on its energy content according to the hydrogen's lower heating value (LHV).
Equation (19) shows the hydrogen tank modeling, while ( 20) is a constraint ensuring that the amount of hydrogen at the end period is equal to or higher than the initial amount.The hydrogen tank outputs are presented based on the level or amount of stored hydrogen, and the conversion is made using (21). 61he formulas for the hydrogen tank in ( 19) consist of various parameters and specifications.E HT represents the value of stored energy in any time step.The amount of stored energy in the hydrogen tank depends on the present energy.Also, the amount of input energy from the electrolyzer and the amount of output power into the fuel cell should be considered to evaluate the amount of stored energy in the hydrogen tank.
The HSS is simulated based on power units (kW).So, all the subsystems have the same unit.If it is needed, the results of the HSS can be transformed to kg based on the hydrogen's LHV.In (21), the stored energy in the hydrogen tank and its unit converting to kg have been presented.
HT HT Hydrogen tank fueling and discharge is one of the main barriers against advances in this area.This particular issue is more visible in HVs that require constant feeling.There are many factors, such as the cost of fueling systems, safety of the tanks, leakage, and so on, that need to be studied and addressed properly.For example, İnci et al. 62 have studied the HV as a storage system in the grid that can participate in power generation as a vehicle-to-grid similar to EVs.The purpose of their study was to enable network integration of hydrogen systems in transportation and electricity production.They developed and implemented a singleswitch high-voltage DC-DC converter to integrate the low-voltage fuel cell to the grid.In another study, the fueling process of HVs was researched. 63A computational fluid dynamic model was used to evaluate the proposed fueling process, along with a flow behavior analysis to study the critical and subcritical processes in the fueling.
Since the efficiency of storing hydrogen is among our focus of study, storage options are also another interesting issue.In future research works, optimal hydrogen storage options, in terms of cost, time, loss, and so forth, are studied incorporating the proposed study.

• Inverter
The inverter modeling is presented in (22), which η INV represents the inverter's efficiency.
( ) • Objective function The OF in this study consists of the following terms: • total net present cost, • initial cost of the energy system, • cost of energy, • cost of emission penalty.
The selected OF minimizes the TNPC, operation and maintenance (O&M) costs, fuel costs, and emission penalties for the MCMG understudy.The optimal design and sizing of each component, such as PV, WT, and CHP, are the optimization problem's decision variables.The mathematical expression of the OF can be presented using (23), 37,64 where NPC C represents the net present costs of MCMG's components and includes the initial, O&M, replacement, and salvage costs as presented in (24).
In (24), N is the project lifetime and ir represents the actual interest rate.The fuel costs are calculated throughout the simulation and are included in the O&M costs.Also, C I , C O&M , and C Rep are the initial, O&M, and replacement costs for each component.These costs are calculated using ( 25)-( 27) 65 and. 32In (26), PWA represents the present worth analysis and is calculated using (28).The actual interest rate is evaluated according to (29), where ir nom and f represent the nominal interest rate and inflation rate, respectively. 24 The COE, renewable penetration, and salvage cost are determined using ( 30)-( 32), 20,32 where S is the salvage cost at the end of the simulation.R rem and R equ represent the remaining and total lifetimes of the component, respectively.In addition, f ren , E ser , E ren , and E tot represent the renewable fraction, the total amount of supplied energy, the amount of renewable energy generation, and the total energy production.
Rep rem equ (32)   The proposed optimization problem aims to find the optimal size of each component.Therefore, the capacities of PV, WT, electrolyzer, hydrogen tank, fuel cell, lithiumion battery, and CHP are the optimization variables.These optimization variables are one-dimensional, meaning that each of them is a 1 × 1 array.Furthermore, the optimal operation of relevant components during each time step of the day are other optimization variables.The charge and discharge power of the storage components and the operational capacity of generators and energy sources should be optimized.Each of these variables is a 1 × 24 array for each day of the simulation.Depending on the time step for optimal operation of the MCMGs, the dimensions of the optimization variables regarding the operation decisions change.
In addition to technical explanations and the concept of the proposed optimization problem and its OF, it is necessary to discuss the performance that guarantees the validity and accuracy of the proposed method.A large number of variables and nonconvexity of the problem make it susceptible to being trapped in local minima and highdimensionality issues.This issue emphasizes the need for guarantees for finding global optima in the proposed study.The HOMER Pro software has been used in this study as the optimization tool.This software has been widely utilized in several academic research and industrial projects.Indeed, HOMER is a well-known software for solving nonconvex optimization problems in energy systems.The advantages of HOMER are highlighted when evaluating a large number of variables and high-dimensionality nature. 66,67t should be noted that the HOMER Pro software utilizes two algorithms to solve optimization problems.The first algorithm is the grid search algorithm, which is used to simulate all possible solutions and combinations of energy sources.The search space defines the feasible configurations, and the algorithm finds the solutions with the minimum cost.Although the grid search algorithm is not complicated, it is a powerful technique for solving optimization problems with multidimensional and multivariable solution spaces. 68nother algorithm used in HOMER Pro is a proprietary derivative-free algorithm.This optimization algorithm is a computational method that solves optimization problems without relying on derivative information.It is particularly useful when computable derivatives are not available or when derivative information is not accessible. 69These optimization algorithms are commonly used in various cases and problems where conventional algorithms may not be effective.Additionally, these algorithms are suitable for cases where designers and providers wish to keep their techniques confidential.While the above-mentioned optimization algorithms are suitable for solving optimization problems for MGs and MCMGs, it is important to conduct additional examinations and evaluations to ensure the performance of the proposed study.This can include sensitivity analyses, statistical analyses on the obtained results, and other relevant evaluations.
In the concept of energy modeling and optimization, artificial intelligence (AI) optimization methods are another approach that has been effectively used to determine the optimal size of energy systems.Kanase-Patil et al. 70 have done review research on AI-based optimization algorithms that have been used for technical and economic studies of integrated renewable energy systems.The algorithms that were discussed in their review included PSO, artificial neural network (ANN), fuzzy logic, adaptive neurofuzzy-based system, radial basic function network, genetic algorithm (GA), and so forth.The results of their review showed that different algorithms have different advantages and disadvantages.For example, swarm optimization algorithms, such as PSO, have better performance in terms of computational time and accuracy, while the GA algorithm needs more calculation time as a result of different solutions existing in each iteration.
Similar review research was conducted by Abdolrasol et al. 71 that studied different AI-based optimization methods with emphasis on the ANNs methods' importance, limitations, and benefits.The study concluded with enhancement solutions to improve the performance of such methods.Solutions that can optimize the parameters involved, like, weights, bias and learning rates, number of hidden layers, number of nodes, and activation functions.The review included test case studies to evaluate the improvement of hybrid ANN optimization methods such as ANN-PSO and ANN-GA and compare their performances in the case of training performance, time, regression, and so forth.Overall, it can be concluded that the AI-based optimization methods can still improve to solve problems in a shorter time and reach global optima.

| TEST RESULTS AND DISCUSSIONS
The proposed reliability-oriented design was implemented on a real MCMG located on the campus of the University of Kashan, in Kashan, Iran.The case study focuses on the arid climate conditions of the desert area of Iran.The University of Kashan and the MCMG under study are geographically located at 36°5.0′ North and 115°9.2′East, situated in the BWh climate zone according to the Koppen-Geiger climate classification system. 72he application of the proposed design to a university campus as an MCMG and the assessment of various technologies for ESSs are significant contributions.HOMER Pro software was utilized for this study.The objective of this study is to examine the impacts of hydrogen and electrical ESSs on the reliability-oriented optimal planning and operation of MCMGs.Average weather data, including solar irradiation, clearness index, and wind speed, were obtained from the NASA Prediction of Worldwide Energy Resources database, 73 as depicted in Figure 2. Realistic hourly solar data was generated using the latitude and monthly average values based on the method proposed by Graham et al. 74 and Graham and Hollands, 75 which is integrated into the synthetic data generation of HOMER Pro software, as the proposed simulation is conducted in hourly time steps.The described data-synthesis algorithm also generates realistic synthetic hourly wind speeds and solar irradiations.Figure 3 illustrates the hourly thermal and electrical demands, representing a typical university building at the University of Kashan with an approximate area of 7000 m 2 .While there is not a significant difference in energy consumption, the average annual energy consumption per-unit area of university campuses depends on various parameters, including but not limited to the number of classes, geographic effects, and the campus size. 76This study aims to assess the benefits of electrical and hydrogen ESS technologies for MCMGs in arid climate conditions based on the structures shown in Figure 1.It is important to note that although the case study is for a university campus located in an urban area, the simulation results are also applicable to islanded educational centers/MCMGs.
The study examined the MCMG under the following scenarios: − Scenario 1: The MCMG includes HSS, and emission penalties are considered.− Scenario 2: The MCMG includes BSS, and emission penalties are considered.− Scenario 3: The MCMG includes HSS, and emission penalties are not considered.
− Scenario 4: The MCMG includes BSS, and emission penalties are not considered.
The technoeconomic specifications of the PV panel are presented in Table 2. 58 The WT type/model used in this study is the Eocycle EO20, and Table 3 shows the characteristics and specifications of the WT. 77,78The battery model assumed in this study is the Tesla Powerwall 2.0, and the technical and economic characteristics of the BSS are presented in Table 4. 58,77 F I G U R E 2 Average and hourly solar irradiation and wind speed.

| 2715
The characteristics of CHP and boiler components are presented in detail in Table 5. 23,80 Additionally, Table 6 shows the HSS characteristics and specifications.Similarly, the characteristics of the inverter have been demonstrated in Table 7. 63 Furthermore, the other simulation parameters are presented in Table 8.The average and hourly solar irradiation and wind speed for the MCMG under study are presented in Figure 2. Figure 3 shows the hourly electrical and thermal load profile of the studied MCMG.The electrical and thermal peak loads are 366.66 and 502.68 kW, respectively, while their corresponding peak months are July and January.
The base case studies under various scenarios are conducted for a 1% allowed MCS, and a sensitivity analysis is conducted for the 3%, 5%, 7%, and 9% allowed MCSs.The percentage is relative to the total energy demand of the MCMG.Therefore, the base case simulation is carried out under scenarios that are limited to having a maximum of 1% capacity shortage.This means that the maximum allowed capacity shortage would be 1%.The percentage is calculated based on the total energy demand of the MCMG.In other words, the amount of energy that is not supplied during the simulation divided by the total energy demand of the MCMG should be less than 1%.Otherwise, that solution would not be feasible due to the constraint violation.
The emission penalty for CO 2 is assumed to be 20 $/ton, 81 while the emission penalty for excess NO and SO 2 is 5800 $/ton. 82According to the data shown in T A B L E 2 Technoeconomic specifications of PV units. 58

Item
Parameter Value  The search space for PV, WT, and BSS is defined via lower and upper bounds as presented in associated characteristics and specifications.In Tables 2 and 3, the associated specifications for PV, WT, and BSS have been introduced, which are used to determine the search space.Also, the search space values for other components, such as CHP, fuel cell, and electrolyzer, are assumed to vary from 0 to 500 kW with a 25-kW step size, while the search space value for hydrogen tank is between 0 and 500 kg with a 25-kg step size.
Sometimes, the lack of consistency of load variation makes it difficult to judge the robustness of the proposed scheme.Using the per-unit values could be an appropriate solution to overcome this challenge.However, in this paper, the actual load profiles have been used.
Since the HOMER optimizer does not allow any loss of thermal load, using the boiler component as a backup generator is inevitable in all simulations.Hence the thermal reliability in all cases and scenarios is at 100%, and the capacity of the boiler's output power (kW) does not have an upper bound.
Table 9 shows the simulation results for 1% MCS under each scenario.These results include the optimal capacity of each component of the studied MCMG and the total CO 2 emissions for each ESS technology.Also, T A B L E 5 Technoeconomic specifications of the CHP and boiler. 23,80em | 2717 the results of optimal operation of CHP and boiler components, such as fuel cost and excess thermal energy generation, have been presented and compared.
It should be noted that at the end of the simulation by HOMER, several optimal answers are provided for different system designs based on the previously mentioned criteria (COE, net present cost, renewable penetration, fuel cost, etc.).The solutions that are presented for each scenario and reliability index in this article are the ones that have the least TNPC.The profitability of the business related to this study is one of the essential issues.Indeed, the best TNPC means that most profitability can be achieved through the business or activity of the MCMG.Integrating the proposed study and other profitability analyses for different businesses is one of the future works.
The data in Table 9 represents the optimal capacity of the hydrogen tank, indicating the amount of hydrogen it can hold, without consideration for the tank's weight.
The emission penalties included in Table 9 are factored into the TNPC and are not deducted.The excess thermal power refers to the percentage of exhaust thermal power of CHP relative to the total thermal power generation of the energy system.The results presented in Table 9 suggest that the TNPC of islanded MCMG with an HSS is higher than the system with the BSS under Scenarios 1 and 2, considering the emission penalties.
Regardless of different scenarios, due to the characteristics of HSS, the MCMG equipped with it has higher renewable penetration and lower CO 2 emissions than the MCMG equipped with electrical-based ESS.It is concluded that, neglecting the emission penalties (Scenarios 3 and 4), the HSS is slightly more economical than the electrical one, while also resulting in lower CO 2 emissions and higher renewable penetration.The application of emission taxes has reduced the system's fuel consumption and cost, leading to higher renewable penetration and lower CO 2 emissions.Comparing the effects of emission penalties on HSS scenarios (1 and 3), it is implied that with a 7.97% increase in TNPC, CO 2 emissions have decreased by 26%, which is significant.Under scenarios using the BSS, the TNPC difference between Scenarios 2 and 4 is around 5.42%, while the emission reduction is 12.96%.This result indicates that even though HSS is more expensive, investing in HSS is more effective than BSS in reducing CO 2 emissions.These results also indicate that considering the emission taxes and transitioning towards cleaner power generation with higher renewable penetration is slightly cheaper if the MCMG is equipped with BSS.The application of emission penalties significantly affects CHP's operation, with a considerable difference in excess thermal power generation.Due to low fuel prices in Iran and similar Middle Eastern countries, without emission taxes, it is more cost-efficient for CHP to supply a considerable portion of electrical demand, whether there are thermal demands or not.This results in significant thermal energy waste, fuel consumption, and CO 2 emissions.According to the results of fuel costs and excess thermal power, the CHP will operate more efficiently by considering emission taxes, resulting in lower levels of exhaust heating power in both technologies of ESSs.Another essential factor influencing excess thermal power generation is related to the nature of the MCMG's application and the electrical and thermal load profiles.The boiler is responsible for supplying adequate power whenever there are only thermal demands.When the electrical demand exceeds the total capacity of renewable energy resources and ESSs, it is inevitable for CHP to supply the electrical demand, regardless of thermal load presence, resulting in 60% recovered heat going to waste if necessary.In summary, the CHP system operates more flexibly under Scenarios 1 and 2, and the emission penalties affect its optimal operation.The study found that the excess thermal power generation is reduced to 44.3% from 81.6% in the MCMG equipped with HSS.The hourly output power of the MCMG components and capacity shortage under each scenario are illustrated in Figures 4 and 5, showing the hourly output power of renewable energy units (PV and WT units) under various scenarios.
As shown in Figure 4, the size of PV is different under multiple scenarios.Test results shown in Figure 4 infer that the maximum PV size is related to Scenario 1. Also, since the solar irradiance would be similar for all scenarios, the pattern of the PV output power should be the same.Only, the maximum output power of PVs under multiple scenarios changes.
The analysis of the test results presented in Figure 5 suggests that the size of the WT under Scenarios 1 and 2 would be higher than in Scenarios 3 and 4. Similar to the PV output power under multiple scenarios, the WT output power under multiple scenarios follows a similar pattern, mainly due to the consistent wind speed across scenarios.However, the higher values for the WT size led to a greater amount of WT output powers under some scenarios.
When comparing the renewable capacities of each scenario, it is evident that the first and second scenarios have higher renewable penetration, highlighting the effectiveness of emission taxes.Figures 6 and 7 depict the output power of the CHP and boiler, respectively.The CHP is utilized for the load-following dispatch strategy, meaning that whenever there is an electrical demand, the CHP operates at an optimal capacity sufficient only to meet the demand, and its output power never charges any ESSs or supplies deferrable loads.The alternative approach is cycle-charging, where any supply unit operates at maximum capacity, and the excess electricity is used to charge the ESSs or supply deferrable loads.
Both strategies (load-following and cycle-charging) have been studied, and test results indicated that regardless of considering the emission taxes, the loadfollowing strategy reduces fuel consumption and operation costs compared with the cycle-charging strategy.Consequently, lower TNPC and CO 2 emissions occur with the load-following strategy.| 2719 The CHP is assumed to be the primary thermal power source in the proposed system, while the boiler serves as a backup heat source.If there is thermal demand without any electrical demand or if the thermal demand of a specific time step is higher than the recovered heat from the CHP based on its electrical output, the boiler is responsible for meeting the total or remaining thermal demand to maintain the thermal reliability of the system.Additionally, in time steps with only thermal demand, the CHP might start to operate if it is economical, and the excess electrical power will be stored in the HSS.Since no emission taxes are considered under Scenarios 3 and 4, the capacity of CHP has increased in these scenarios.Moreover, the optimal design of a reliable MCMG based on BSS requires a higher capacity of CHP, resulting in lower renewable penetration for the BSS-based MCMG compared with the HSS-based one.
Test results in Figure 6 show that the electric and heat output power of CHP would differ under multiple scenarios.Test results show that the minimum electric output power of CHP is needed under Scenario 1.The output power of the CHP depends on the amount of available power for PVs and WTs.The heat output power of the CHP is also different under multiple scenarios.If the CHP provides the electric output power, the heat power is also produced.The technologies for the ESS will affect the amount of power of the CHP.
The analysis of the test results presented in Figure 7 indicates that the maximum power output from the boiler does not vary significantly across scenarios.Additionally, the pattern of the boiler's output power under multiple scenarios appears to be similar.By comparing the test results shown in Figures 3, 6, and 7, it can be concluded that there is a correlation between the boiler's output power and the thermal demand profiles.During peak electrical demand periods, the majority of the thermal demand is met through the heat recovery system of the CHP, resulting in minimal noticeable output power from the boiler.Conversely, during periods of peak thermal demand and lower electrical demand, the boiler must complement the CHP's thermal output power to meet the required demand.
Figure 8 illustrates the charging scheduling and SOC of the BSS under Scenarios 2 and 4. A comparison of the results under Scenarios 2 and 4, as shown in Table 9, suggests that the capacity of the BSS has low sensitivity to emission taxes, while the MCMG is expected to provide high-reliability electrical power.The test results in Table 9 indicate that under Scenarios 2 and 4, approximately 11 and 12 battery units have been assigned to the MCMG, respectively, which are nearly identical.Figure 8A,C depicts the BSS scheduling under Scenarios 2 and 4, showing a similar profile and pattern optimized for the BSS under both scenarios.In this study's simulations, thermal reliability has been assumed to be consistently at 100% due to the presence of the boiler.The system's performance for different electrical power reliability values has been evaluated.Therefore, when highly reliable conditions are addressed in the test results and discussions, it refers to highly reliable electrical power/electricity supply.
The optimization results of the HSS, which includes the fuel cell, electrolyzer, and hydrogen tank, under Scenarios 1 and 3 are depicted in Figure 9.The test results under Scenarios 1 and 3 reveal a direct correlation between the power capacities of the fuel cell and electrolyzer and the energy capacity of the hydrogen tank (kg) due to their operational dependency.It is observed from the test results in Figure 9 under Scenarios 1 and 3 that a higher storage capacity of the hydrogen tank is required if a higher capacity is assigned to the fuel cell and electrolyzer.Comparing the test results under Scenarios 1 and 3 based on Table 9 and Figure 9 highlights the correlation between the size of the hydrogen systems.It has been observed that the storage capacity increases when the BSS is used under Scenario 4 compared with Scenario 2. On the contrary, while the HSS is used, the capacity of the ESS decreases significantly under Scenario 3 than Scenario 1.The test results also indicate that considering emission taxes increases the storage capacity of the HSS, while the battery system decreases by one unit.The HSS has lower emissions in both scenarios.Hence, it has been concluded that the CHP operates in a way that the MCMG requires more storage to maintain a high level of power supply reliability.The required storage capacity will be higher compared with the CHP system's operation in a battery-based MCMG.Decreasing the output power of the CHP would be a more cost-efficient solution when emission penalties are considered.In these cases, using the ESS would be a better solution.On the other hand, for the BSS-based MCMG, it would be more beneficial (cost-wise) for the CHP to run more often, even if there are emission penalties involved.It is worth mentioning that the optimal capacity of the MCMG components is highly dependent on various factors, such as the operation of the renewable sources that could cause a different peak of power for charging the storage system, operation of CHP, and load profile.Therefore, it would be difficult to conclude the exact reason behind a specific result without studying subsystems' operation and hourly results in detail, which can be a tremendous future work for this study.
However, there is a similar trend in the unmet electrical load in both technologies of ESS (electrical and hydrogen technologies).The hourly capacity shortages under multiple scenarios are presented in Figure 10.Test results regarding capacity shortages infer that the impacts of emission penalties on the unmet electrical load are more significant when the BSS technology is utilized.In comparison, the capacity shortage difference in the HSS-based MCMG is less than 4%.Moreover, the comparison between the unmet electrical load of each scenario indicates that the amount of capacity shortage and its frequency/probability in the BSS-based MCMG is lower than in the hydrogen-based one.It can be concluded that without the emission penalties, the HSS is better than the BSS in 99% power supply reliability.It would be better from the aspects of cost, renewable fraction, and emission.So, the HSS is overall a better system than the BSS in base case studies.Indeed, the difference between the capacity shortage of the two ESS technologies is more than 64%, while the emission penalties are not considered.It should be clarified that the above conclusions are not overall conclusions for all of the systems.Indeed, the technoeconomic factors and results are dependent on many determining factors in a multivariable simulation like this.Hence, it cannot be concluded for 100% why the system's load loss is greater in the HSS-based MCMG in both scenarios.It could be because of emissions, cost of components, cost of operations, or every operation of components.For this study and assumed conditions, it has resulted that the BSS has less load loss than HSS.Therefore, for similar systems, similar results could appear, while checking the similarity of other case studies with this test system is necessary.
The CHP plays an essential role in the yearly capacity shortage.The CHP can operate as a backup generator if electrical demand exceeds the available output power of renewable energy sources and ESSs, particularly in peak hours.If a higher capacity is assigned to the MCMG, the system reliability can be improved.On the other hand, other essential issues should be considered for sizing the CHP.However, since the optimal operation of CHP is affected by many decision variables, such as the output power of renewable energy sources, the emission penalty constraints, MCS allowed, and the optimal capacity of CHP, the results under different scenarios do not follow the same trend.For instance, in both scenarios related to HSS-based MCMG (Scenarios 1 and 3), the optimal capacity of the CHP is less than the corresponding scenario of the BSS-MCMG (Scenarios 2 and 4), This is mainly one of the reasons behind the higher power capacity shortage when the HSS technology is utilized.To get insight into how the allowed MCS affects the proposed reliability-oriented optimal design of the MCMG using different ESS technologies (electrical and hydrogen technologies), sensitivity analyses are performed.The MCMG under study is analyzed, while 3%, 5%, and 9% MCS are considered in the proposed reliability-oriented design.In addition to the base case, in which MCS has been assumed to be 1%, other reliability cases have been studied.Different values for the allowed MCS and load loss for the MCMG have been examined in this paper.The system's reliability is 97% for the 3% MCS.It means the system should (is constrained to) supply 97% of the total energy demand and it can only have load loss (in kWh) up to 3% of the total energy demand per year.
Table 10 depicts the simulation results for a 3% MCS, showing that regardless of the scenarios (considering or neglecting emission penalties), the TNPC of an MCMG | 2725 equipped with HSS would be higher than a BSS-based one.However, the fuel cost/consumption in the HSSbased MCMG and the CO 2 emissions are lower than the BSS-based one under all scenarios.One of the key points from the results is the impact of emission penalties.When considering emission penalties for the HSS-based MCMG (comparing Scenarios 1 and 3), the TNPC increases by 6.98% ($180,829).On the other hand, CO 2 emissions decreased by 18.25% (42,431 kg/year), which could be considered very interesting a policymaker's point of view.The same comparison for the BSSbased MCMG resulted in a 9.2% increase in TNPC ($236,935).Moreover, by applying emission penalties for the BSS-based MCMG, a 5.88% decrease in CO 2 emissions (21,900 kg/year) has been obtained.
For the HSS-based MCMG, not only is the difference in CO 2 reduction greater than for the BSS-based one, but also the cost of transitioning towards a cleaner system (applying emission taxes for already established systems) is more significant.In addition, the TNPC difference between Scenarios 1 and 3 (with and without emission penalties) for HSS-based MCMG is 6.98% ($180,829), which is less than for the BSS-based MCMG.Indeed, the BSS-based MCMG has a TNPC difference of 9.2% ($236,935).So, if it is assumed to update the MCMG design based on the emission taxes in the modeling, the difference in cost is less when using HSS.As seen in Table 10, the HSS-based MCMG has less excess thermal energy compared with the BSS-based one in both scenarios of emission penalties (with and without).By comparing Scenarios 1 and 3 for the HSS-based MCMG and Scenarios 2 and 4 for the BSS-based MCMG, it can be concluded that the HSS-based MCMG has less thermal waste than the BSS-based one.This result indicates that the CHP and boiler can operate more optimally and flexibly in HSS-based MCMGs, particularly when the energy system is subjected to emission penalties.
The results indicate that the cost of emission penalties for a 3% MCS is similar to the base case (1% MCS).However, there is a significant difference in the cost of emission penalties between ESS technologies.For the base case (1% MCS), the emission penalty for both storage technologies is nearly the same.But for 3% MCS, the BSS-based MCMG has an 8.83% higher emission penalty than the HSS-based one.Moreover, the TNPC difference between Scenarios 1 and 2 has reduced significantly.The TNPC difference between the first two scenarios is only $14,753, with a $18,915 difference between their emission taxes.This result indicates that in the presence of emission taxes, it is more economical to employ BSS-based MGs instead of HSS.However, an HSS-based MCMG has lower emission taxes.From an environmental aspect, an HSS-based MCMG will reduce CO 2 emissions by 17.5% due to its higher renewable fraction and lower excess thermal energy production with only a 0.56% increase in the system's TNPC.
Furthermore, without environmental limitations at this reliability level, the BSS-based MCMG is the cost-efficient alternative for high-reliability energy systems.Comparing the TNPC under multiple scenarios and also the differences between the renewable fractions of different scenarios could be useful to analyze different designs and technologies for the MCMGs.The comparison of TNPC and renewable fractions under multiple scenarios suggests that HSSbased MCMGs are much more environmentally friendly than BSS-based ones.In proper conditions and applications, HSS-based MCMGs can be more beneficial, even though their investment cost might be higher.
It is observed that applying emission penalties in the optimization had no influence on the capacities of HSS, and the optimal capacity of the fuel cell, electrolyzer, and hydrogen tank has remained the same in both scenarios.In contrast, implementing emission taxes has increased the BSS's optimal capacity.Moreover, the CHP capacity has increased in Scenario 3 since its operation has no limitations from an environmental aspect and is cheaper to utilize.For the battery system, the optimal capacity of CHP remained the same, and only the amount of fuel consumption (comparing the fuel costs) increased.This is also true for HSS-based MCMG, where both the CHP and fuel consumption have increased.
The hourly scheduling and operation of the MCMG's subsystem and capacity shortage under different scenarios of 3% MCS are presented in Figures 11 and 12.As shown in Figure 11, the capacity of BSS has increased in Scenario 2. However, the optimal capacity of BSS still has low sensitivity to emission tax (Comparison of  F I G U R E 12 Optimal hydrogen tank level of the HSS under Scenarios 1 and 3, while MCS is assumed to be 3%.(A) Hydrogen tank level under Scenario 1 and (B) hydrogen tank level under Scenario 3. HSS, hydrogen storage system; MCS, maximum capacity shortage.
Scenarios 2 and 4). Figure 12 and the obtained results show that the capacity of HSS components has not changed in Scenarios 1 and 3, and the only difference is in the optimal operation of corresponding components, that is, fuel cell, electrolyzer, and hydrogen tank.In addition, Figure 13 depicts the hourly capacity shortage corresponding to a 3% allowed MCS under multiple scenarios.
The yearly capacity shortage of the BSS is lower than HSS in all scenarios, and the peak of load loss in the first scenario has a higher value than in the second scenario.
Increasing the MCS from 1% to 3% has allowed the MCMG to have 2% more load loss during the year.This increase includes an increase in load loss occurrence (the number of load loss happened during the year) and the load loss amount.The results provided do not show the increase in the occurrences, but the increase in amount can be calculated using the yearly load loss provided in Tables 9 and 10.The increases in load loss due to 3% MCS compared with 1% MCS under Scenarios 1-4 are 19,636, 17,375, 19,543, and 23,636 kWh/year, respectively.Moreover, the number of load loss conditions has been extracted from the results.For 1% MCS, 263, 268, 313, and 101 would be the number of load loss conditions under Scenarios 1-4, respectively.If 3% MCS is considered to design the MCMG, the number of load loss increases to 694, 652, 648, and 612 under Scenarios 1-4.
The results of the optimal design and operation of HSS and BSS-based MCMG in arid climate conditions for 5% MCS are presented in Table 11.
As revealed by test results shown in Table 11, the decrease in the MCMG's reliability reduces the TNPC difference between the two technologies.In this reliability level (MCS = 5%), the TNPC difference between Scenarios 3 and 4 is only 0.1% ($2373) in favor of the HSS-based MCMG.The fuel cost/consumption and CO 2 emissions are also less in HSS-based MCMG, regardless of the emissions penalties.The HSS-based MCMG also has a higher renewable fraction and lower excess thermal power, while the emissions penalty in both MCMGs is nearly the same.Comparing the first two scenarios, the TNPC difference between them is 2.66% (65,902), and the difference between their CO 2 emissions is 6.59%.This result indicates that with only a 2.66% increase in TNPC in HSS-based MCMG, the CO 2 emission has been reduced by 6.59%.The result of the same comparison for the third and fourth scenarios shows that along with a 0.1% less TNPC in HSS, the CO 2 emissions are reduced I G U R E 13 MCMG capacity shortage under multiple scenarios, while MCS is assumed to be 3%.(A) Scenario 1, (B) Scenario 2, (C) Scenario 3, and (D) Scenario 4. MCMG, multicarrier microgrid; MCS, maximum capacity shortage.by 12.45% as well.Although the HSS-based MCMG in this reliability level is still more expensive than BSS ones when emissions penalties are involved, their significant difference in emissions level outweighs their price from an environmental aspect.Similar to other reliabilities levels, from the economic aspect and without any environmental concerns, the HSS-based MCMG is a cost-efficient alternative for university campuses to utilize, while they also have various environmental benefits.
The results also show that the optimal capacity of CHP is the same in all scenarios.Regardless of the previous reliability levels that CHP's capacity influences the simulation results, the optimal CHP operation is the only determining factor affecting the emissions penalty and CO 2 emissions.The emissions taxes have reduced the excess thermal power and increased the MCMG's renewable fraction under Scenarios 1 and 2. Comparing the results of considering emissions taxes, the TNPC of the HSS-based MCMG has increased by 9.46%, while the CO2 emissions have dropped by 4.39%.The results of the same comparison for the BSS-based MCMG are 7.01% and 10.39%, respectively.These results illustrate that considering emissions taxes for educational applications has better economic and environmental impacts on the BSS-based MCMG where this energy system has lower TNPC difference and considerable emissions reduction.Moreover, the storage capacity of both cases has decreased in this section.
Applying the emissions taxes did not affect the optimal design (capacity) of HSS components based on Scenarios 1 and 3.However, considering emissions penalties has increased the BSS capacity significantly (from 3 to 8 units) under Scenarios 2 and 4.
The hourly results of HSS and BSS components corresponding to 5% allowed MCS are presented in Figures 14 and 15.
It is apparent that the optimal capacity of BSS has been affected significantly in the presence of emissions penalties under Scenarios 2 and 4. The test results shown in Table 11 could be useful for comparing the results.Also, the charge/discharge power of the BSS in Figure 14 infers that a significant difference exists between Scenarios 2 and 4 for the BSS-based MCMG.
Moreover, it appears the second scenario has a relatively higher SOC during the year than the fourth  F I G U R E 15 Optimal hydrogen tank level of the HSS under Scenarios 1 and 3, while MCS is assumed to be 5%.(A) Hydrogen tank level under Scenario 1 and (B) hydrogen tank level under Scenario 3. HSS, hydrogen storage system; MCS, maximum capacity shortage.
scenario.Figure 15 depicts the hydrogen tank level under Scenarios 1 and 3, while 5% MCS has been considered.It is apparent that the hydrogen tank has the same size in both scenarios and the difference is in the operation hours and amount of charge and discharge.Also, it can be seen that the storage tank in both scenarios follows a similar trend throughout the year.
The size of the HSS components (hydrogen tank, electrolyzer, and fuel cell) is the same under Scenarios 1 and 3.However, the hydrogen tank level under Scenario 1 appears to have higher values during the year than in Scenario 3. Also, the hourly capacity shortage corresponding to 5% MCS is presented in Figure 16.
On the basis of the results presented in Table 11, it is evident that Scenario 1 has a lower unmet electrical load compared with Scenario 2. However, without emissions penalties, the results are reversed.When comparing the BSS scenarios, the inclusion of emissions taxes has led to an increase in yearly load loss in Scenario 2. On the other hand, the load loss has decreased significantly in the HSS scenario due to the emissions penalties.Additionally, Figure 16 complements the results shown in Table 11, showing similar patterns of unmet electrical loads under multiple scenarios for 5% MCS.This suggests that applying reliability limits, such as MCS, would lead to a more reliable MCMG design regardless of the ESS technologies.The results of the optimal design of each component in the MCMG with a 7% MCS, consisting of HSS or BSS storage systems, are presented in Table 12.It is observed that in the 7% MCS, the difference in TNPC between the two technologies has decreased further, indicating that in an MCMG with lower sensitivity to load loss, HSS and BSS are nearly equal from an investment perspective.Specifically, for scenarios with emissions penalties (Scenarios 1 and 2), the BSS-based MCMG has a lower TNPC compared with the HSS-based one.The results for 7% MCS demonstrate that the choice of ESS technology is influenced by higher allowed MCS values.Unlike 1%, 3%, and 5% MCS, it has been observed that the HSS-based MCMG has a lower TNPC when comparing the first two scenarios, while maintaining an advantageous CO 2 emissions level.
When comparing the results of the first two scenarios, it is evident that the HSS is 1.52% ($36,102) less than the BSS to invest in.Additionally, the HSS-based MCMG has a higher renewable fraction, lower fuel cost, and lower emissions penalty compared with BSS.The difference in CO 2 emissions between the two technologies is 18.17% in favor of the HSS.Furthermore, the HSS-based MCMG F I G U R E 16 MCMG capacity shortage under multiple scenarios, while MCS is assumed to be 5%.has lower excess thermal energy generation and slightly lower load loss during the year.Contrary to previous reliability levels, where there are advantages and disadvantages in utilizing HSS or BSS in an MCMG from either economic or environmental aspects, the HSS in the first two scenarios is ultimately advantageous against BSS comprehensively.Moreover, the optimal capacity of CHP in the HSS-based MCMG is lower than BSS in all scenarios, resulting in a total fuel cost that is 9.22% and 14.14% less in Scenarios 1 and 3, respectively, compared with Scenarios 2 and 4. In addition to lower emissions and fuel costs and higher renewable penetration, the HSS also has a lower load loss during the year in the first two scenarios.However, the simulation results are slightly different under Scenarios 3 and 4. Without emissions penalties, BSS has a lower TNPC than HSS, while HSS still has lower CO 2 emissions.The difference between the TNPC and CO 2 emissions of these scenarios is 1.1% ($23,726) and 25.69%, respectively.It is concluded that by utilizing an HSS-based MCMG, even if there are no emissions restrictions, a 25.69% emissions reduction is achieved with only a 1.1% increase in investment cost.Furthermore, the HSS-based MCMG has a higher renewable fraction, lower fuel cost, and lower excess thermal energy generation.
The difference between yearly excess thermal power generation is very noticeable in this section, especially when there are no emissions penalties and no environmental limitations in using the CHP component.The results show that the implementation of emissions taxes did not affect the HSS's component optimal capacity while it caused a considerable increase in BSS capacity.In the presence of emissions penalties, the comparison between the scenarios of BSS shows a 9.85% increase in TNPC and a 2.37% reduction in CO 2 emissions.Moreover, implementing the emissions taxes has also reduced the fuel cost and thermal energy waste and increased the system's renewable fraction in all scenarios.
Figures 17 and 18 show the optimized operation and scheduling of BSS and HSS corresponding to 7% MCS under different scenarios, respectively.Test results infer that the BSS's optimal capacity has increased significantly in Scenario 2, corresponding to 7% MCS.Additionally, the SOC in Scenario 4 has relatively lower values during the year, and the amount of the BSS's preserved energy is higher in Scenario 2. It can be concluded that the battery storage capacity has increased in the second scenario compared with the fourth scenario.Also, the load loss has slightly increased when the emissions penalties are applied.Furthermore, Figure 18 shows the optimal operation and scheduling of MCMG's hydrogen tank when the allowed MCS is assumed to be 7%.The inclusion of emissions penalties in simulations has no effect on the optimal capacities of HSS components, that is, hydrogen tank, electrolyzer, and fuel cell.Moreover, the capacity shortage of MCMG understudy, corresponding to 7% allowed MCS under multiple scenarios are shown in Figure 19.
According to the presented results in Table 12, Scenario 1 has lower unmet electrical demand than Scenario 2. The result of the same comparison is vice versa in Scenarios 3 and 4, and the HSS-based MCMG has a higher unmet electrical load than the BSS-based MCMG, while 7% allowed MCS is considered.The consideration of emissions taxes in HSS has improved power supply reliability, while BSS has a higher capacity shortage in Scenario 2.
The overall simulation results of each scenario, while 9% MCS is considered for MCMG understudy, are presented in Table 13.
The difference in TNPC and CO 2 emissions between the first two scenarios at a 9% MCS is 0.35% and 0.58%, respectively.When comparing the first two scenarios, the HSS-based MCMG has lower TNPC and CO 2 emissions.However, without emission penalties, the BSS-based MCMG has lower TNPC, while the HSS-based MCMG still remains the cleaner energy system with lower CO 2 emissions.The TNPC difference between the second two scenarios is 3.72%, with an 8.92% difference in their CO 2 emissions.The results indicate that considering emission taxes has significantly reduced emissions.Since the MG in all scenarios has the same capacity as CHP, the optimal operation of CHP alongside the corresponding ESS plays an influential role in the amount of CO 2 emissions, fuel cost, and excess thermal energy generation.It is clear that without any emission limitations, there is only a slight difference between the TNPC of both ESS technologies.However, in all system reliability F I G U R E 19 MCMG capacity shortage under multiple scenarios while MCS is assumed to be 7%.| 2735 levels, HSS scenarios have lower CO 2 emissions.The TNPC difference between the HSS-based and BSS-based MCMGs in the first two scenarios continued to decrease with power supply reliability, and according to the 9% MCS results, the TNPC of the HSS-based MCMG has a lower value than the BSS one.The results from Table 13 show that the first scenario has a lower TNPC than the second scenario.These results indicate that in MCMGs requiring lower reliability, which are less sensitive to the capacity shortage, the HSS is more economical to utilize than BSS.It is worth mentioning that even though the TNPC of the HSS is higher than BSS in some scenarios, the HSS-based MCMG has lower CO 2 emissions in all scenarios and reliability levels.
The optimal scheduling and operation of HSS and BSS of the MCMG, corresponding to 9% MCS, are shown in Figures 20 and 21, respectively.The results show that the optimum capacity of HSS with and without emissions tax for MCSs equal to or more than 3% remained unchanged at 25 kW for the electrolyzer and fuel cell and 25 kg for the hydrogen tank.Moreover, the capacity shortages of the MCMG also have been presented in Figure 22.Table 13 and Figure 22 illustrate that the unmet electrical load is higher in the HSS-based MCMG for Scenario 1 compared with Scenario 3. When comparing the BSS and HSS scenarios with emissions taxes (Scenarios 1 and 2), it is evident that the HSS has a much higher unmet load than the BSS.Comparing the BSS scenarios (Scenarios 2 and 4), it is apparent that Scenario 2 has significantly less unmet load than Scenario 4. Unlike HSS, where the scenario with emissions taxes (Scenario 1) has a higher unmet load than the one without the emissions tax (Scenario 3).In summary, regardless of reliability concerns and desired MCS, the HSS-based MCMG (Scenarios 1 and 3) has higher environmental benefits than the BSS-based one (Scenarios 2 and 4).Interestingly, the HSS-based MCMG would be a better alternative from an environmental emissions perspective, even though emissions penalties have not been a concern in the proposed method.On the other hand, the HSS-based MCMG may not be an economically attractive alternative compared with BSSbased MCMGs due to the significant investment cost of HSS.Additionally, the integration of CHP and BSS in MCMGs would make the BSS-based MCMGs more  economical.However, for the studied system (a university campus in arid climate conditions), after a certain reliability constraint (93%), the HSS-based MCMG has lower TNPC and is more economical for future investment while still retaining its environmental traits.Furthermore, it is important to note that the CHP and boiler are the primary sources of emissions in the MG.It is evident that when operating alongside HSS, regardless of their optimal capacity and emissions limitations, the CHP and boiler operate more efficiently, resulting in lower CO 2 emissions in HSS-based MGs compared with BSS MGs.This difference in emissions production is mainly attributed to the operation of the CHP and boiler.Additionally, the HSS-based MGs exhibit lower CO 2 emissions and reduced emissions tax costs across all reliability levels and emissions penalty scenarios.Therefore, it can be inferred that the CHP and boiler operate in a more environmentally friendly manner when integrated with HSS.Table 14 provides a summary of the differences in TNPC between Scenarios 1 and 2, based on changes in the allowed MG control strategies (MCS).
Similarly, However, by utilizing the findings of this study, it is possible to determine the best technoeconomic design for the MG based on its reliability requirements.
The comparative test results and discussions with available references would be useful.Hence, in addition to comprehensive studies in this paper, the obtained results are also compared with available research works.Some of the results for similar available studies for component outputs are presented.Afterwards, the obtained results based on the proposed study are shown to illustrate that similar trends or improved ones are achieved.
Zhang et al. 38 have studied the optimal planning of hybrid MGs, including PVs, WTs, and HSSs subjected to different values of LPSP, that is, 0%, 1%, 3%, 5%, and 10%.The results for the operation and schedule of PV, WT, and HSS for Zhang et al., 38 and the comparison of the obtained results based on the proposed method in this article and Zhang et al. 38 are given.This article's studies have been done in different power supply reliability constraints as well as Zhang et al. 38 This paper has studied 1%, 3%, 5%, 7%, and 9% allowable MCS with respect to full load demand.The HSS's SOC in a scenario without emission penalties, while allowable MCS is considered to be 1%, is presented in Figure 23.
It is worth noting that the storage behavior is highly affected by the load profile.In this paper, the demand load profile is for an administrative building at the University of Kashan's campus in arid climate conditions, which affects the capacity shortages.It has also different peak hours and load shapes from other case studies.The results for HSS are given in kg stored hydrogen, which can easily be converted to kWh using hydrogen's LHV like Zhang et al. 38 It should be noted that there are research gaps in available studies, such as Zhang et al. 38 This paper aims to fill such gaps.In this section, it is examined how this study could respond to such research gaps.The first gap is neglecting the effects of emission penalties on storage operations.The results of HSS operation with 1% allowed MCS, considering the emission penalties, are presented in Figure 24.Comparing the results shown in Figure 23 (similar to Zhang et al. 38 ) and Figure 24, based on the proposed study, illustrates the impacts of considering the emission penalties.
Another contribution of this study is the comparison between the HSS technology and rival technologies, such as BSS.This comparison analysis is addressed in this article, comparing HSS and BSS in terms of capital cost, operation cost, emission, and load loss.The following results are for the BSS with a 1% MCS under both scenarios with and without emission penalties.Test results for BSS's SOC without emission penalties are shown in Figure 25.It is worth mentioning that the battery used in this study has a minimum SOC constraint of 20%.This minimum SOC is assigned because of efficiency and life cycle reasons.Figure 26 shows the BSS SOC with 1% MCS, while emission penalties are applied according to the proposed study.
As the effects of different LPSPs on HSS operation were studied in Zhang et al., 38 also the effect of power supply reliability is investigated on storage systems' behavior in this article.However, factors such as the effects of reliability constraints on the optimal design of the storage system, amount of emission, load loss, and renewable penetration of the overall system have been also covered in the proposed study, which has not been mentioned in available references, such as Zhang et al. 38 The results of including these factors in the simulation F I G U R E 23 HSS's SOC for a short period based on the proposed study without considering the emission penalties, while MCS is assumed to be 1%.HSS, hydrogen storage system; MCS, maximum capacity shortage; SOC, state of charge.
F I G U R E 24 HSS's SOC for a short period based on the proposed study considering the emission penalties, while MCS is assumed to be 1%.HSS, hydrogen storage system; MCS, maximum capacity shortage; SOC, state of charge.
F I G U R E 25 BSS's SOC for a short period based on the proposed study without considering the emission penalties, while MCS is assumed to be 1%.BSS, battery storage system; MCS, maximum capacity shortage; SOC, state of charge.
F I G U R E 26 BSS's SOC for a short period based on the proposed study considering the emission penalties, while MCS is assumed to be 1%.BSS, battery storage system; MCS, maximum capacity shortage; SOC, state of charge.
show that in the base case study (1% allowed MCS) the HSS-based MCMG is more expensive than the BSS-based one, when emission penalties are considered.However, without emission penalties, the HSS-based system would be cheaper to employ.Nevertheless, the cost of the HSS under scenarios with emission penalties gets closer to the BSS as the reliability of the MCMG decreases, and after 7% MCS, the HSS-based MCMG proved to be cheaper than the BSS-based MCMG to establish.Meanwhile, the HSS-based MCMG has fewer CO 2 emissions per year than the battery in all of the tested reliabilities, regardless of the presence of emission penalties (in both scenarios with and without emission penalties).

| CONCLUSION
This study evaluates the impacts of electrical and hydrogen ESS technologies on the reliability-oriented optimal design of multicarrier energy MGs, considering the technoeconomic and environmental aspects.To conduct the simulation, an MG at the campus of Kashan University was modeled under four scenarios, considering the two storage systems mentioned earlier and emissions penalties, using Homer Pro software.Moreover, sensitivity analyses were performed to gain insight into how the multicarrier system (MCS) and reliability concerns affect the optimal design and technoeconomic and environmental behaviors of BSSs-and HSSs-based MCMGs.
The results of this study show that enhancing the reliability of MCMGs equipped with electrical and hydrogen ESSs up to 99% has resulted in a 7.97% increase in TNPC and a 26% decrease in CO 2 emissions.The evaluations have also shown that emissions penalties affect the optimal operation of CHP, especially when the MCMG is equipped with HSS, while the excess thermal power generation is reduced to 44.3% from 81.6%.
These findings suggest that if a high level of reliability is expected, the BSS-based MCMG would be more economical.The results of this study also indicate that, if significant interruptions and unmet loads are acceptable, the HSS-based MCMG could be an effective alternative from the viewpoint of technoeconomic and environmental aspects.On the other hand, regardless of the desired reliability levels and emissions penalties, the HSS-based MCMG would be an environmentally friendly choice.This study establishes a quantitative framework for proposing ESS technologies for the optimal energy management of a campus MG in arid climate conditions.Further research could study the impact of advanced ESS integration on MCMG reliability and efficiency of a university campus under grid connection disruptions or emergency situations.The scalability analysis is another important subject that can be done in future studies.It should be noted that conducting experiments to show the effectiveness of the proposed schemes.Since the proposed study has been applied to an actual MCMG, it can be further evaluated in the future.

F I G U R E 1
Structure of the MCMG under study: (A) using hydrogen ESS and (B) using electrical ESS.AC, alternating current; CHP, combined heat and power; DC, direct current; ESS, energy storage system; MCMG, multicarrier microgrid; PV, photovoltaic.IMANLOOZADEH ET AL.
(A) Average daily solar irradiance, (B) average wind speed, (C) hourly global solar irradiance, and (D) hourly wind speed.F I G U R E 3 Hourly electrical and thermal demand.(A) Electrical load demand and (B) thermal load demand.IMANLOOZADEH ET AL.

F
I G U R E 5 WT output power under multiple scenarios in the base case (MCS = 1%).(A) Scenario 1, (B) Scenario 2, (C) Scenario 3, and (D) Scenario 4. MCS, maximum capacity shortage; WT, wind turbine.IMANLOOZADEH ET AL.F I G U R E 6 Electrical and thermal output power of CHP under multiple scenarios in the base case (MCS = 1%).(A) Electrical output power under Scenario 1, (B) electrical output power under Scenario 3, (C) heat output power under Scenario 1, (D) heat output power under Scenario 3, (E) electrical output power under Scenario 2, (F) electrical output power under Scenario 4, (G) heat output power under Scenario 2, and (H) heat output power under Scenario 4. CHP, combined heat and power; MCS, maximum capacity shortage.

I G U R E 8
Optimal scheduling and SOC of the BSS under Scenarios 2 and 4 in the base case (MCS = 1%).(A) Scheduling of the BSS under Scenario 2, (B) SOC of the BSS under Scenario 2, (C) scheduling of the BSS under Scenario 4, and (D) SOC of the BSS under Scenario 4. BSS, battery storage system; MCS, maximum capacity shortage; SOC, state of charge.

F I G U R E 9
Optimal scheduling and SOC of the HSS under Scenarios 1 and 3 in the base case (MCS = 1%).(A) Hydrogen tank level under Scenario 1, (B) hydrogen tank level under Scenario 3, (C) fuel cell scheduling under Scenario 1, (D) fuel cell scheduling under Scenario 3, (E) electrolyzer scheduling under Scenario 1, and (F) electrolyzer scheduling under Scenario 3. HSS, hydrogen storage system; MCS, maximum capacity shortage; SOC, state of charge.IMANLOOZADEH ET AL.

F I G U R E 14
Optimal scheduling and SOC of the BSS under Scenarios 2 and 4, while MCS is assumed to be 5%.(A) Scheduling of the BSS under Scenario 2, (B) SOC of the BSS under Scenario 2, (C) scheduling of the BSS under Scenario 4, and (D) SOC of the BSS under Scenario 4. BSS, battery storage system; MCS, maximum capacity shortage; SOC, state of charge.

F I G U R E 20
Optimal scheduling and SOC of the BSS under Scenarios 2 and 4 in the base case (MCS = 9%).(A) Scheduling of the BSS under Scenario 2, (B) SOC of the BSS under Scenario 2, (C) scheduling of the BSS under Scenario 4, and (D) SOC of the BSS under Scenario 4. BSS, battery storage system; MCS, maximum capacity shortage; SOC, state of charge F I G U R E 21 Optimal hydrogen tank level of the HSS under Scenarios 1 and 3, while MCS is assumed to be 9%.(A) HSS tank level in Scenario 1 and (B) HSS tank level in Scenario 3. HSS, hydrogen storage system; MCS, maximum capacity shortage.
Summary of the literature review in optimal design of MGs and MCMGs.
, there is a research gap when it comes to evaluating the impacts of different ESS technologies, such as BSS and HSS, on the reliability-oriented design of MCMGs.Another gap in the literature is the lack of studies that focus on the environmental concerns associated with various ESS technologies, including BSS and HSS.The majority of the available references primarily concentrate on electrical loads in MGs, with less attention given to MCMGs and the simultaneous integrated management of heat and electrical demands.

Table 8 ,
the simulation is conducted for a project's lifetime of 25 years.The fuel prices for diesel and natural gas fuels are relatively low in Iran.Thus, the export prices of fuels are used in this article.
77,61gen ESS characteristics.38,61Invertercharacteristics.77 Abbreviations: CHP, combined heat and power; IC, initial cost; O&M, operation and maintenance; RC, replacement cost.T A B L E 6Abbreviations: ESS, energy storage system; IC, initial cost; O&M, operation and maintenance; RC, replacement cost.T A B L E 7 Technoeconomic specifications of fuels and other economic data of multicarrier microgrid under study.
Results of optimal design of studied MCMG for MCS = 1% under various scenarios.
T A B L E 10 Results of optimal design of studied MCMG for MCS = 3% under multiple scenarios.
T A B L E 11 Results of optimal design of studied MCMG for MCS = 5% under multiple scenarios.
T A B L E 12 Results of optimal design of studied MCMG for MCS = 7% under multiple scenarios.
Abbreviations: CHP, combined heat and power; COE, cost of energy; MCMG, multicarrier microgrid; MCS, maximum capacity shortage; O&M, operation and maintenance; PV, photovoltaic; TNPC, total net present cost; WT, wind turbine.IMANLOOZADEH ET AL.F I G U R E 17 Optimal scheduling and SOC of the BSS under Scenarios 2 and 4, while MCS is assumed to be 7%.(A)Scheduling of the BSS under Scenario 2, (B) SOC of the BSS under Scenario 2, (C) scheduling of the BSS under Scenario 4, and (D) SOC of the BSS under Scenario 4. BSS, battery storage system; MCS, maximum capacity shortage; SOC, state of charge.F I G U R E 18 Optimal hydrogen tank level of the HSS under Scenarios 1 and 3, while MCS is assumed to be 7%.(A) Hydrogen tank level under Scenario 1 and (B) hydrogen tank level under Scenario 3. HSS, hydrogen storage system; MCS, maximum capacity shortage.
T A B L E 13 Results of optimal design of studied MCMG for MCS = 9% under multiple scenarios.
Table 14 also demonstrates the TNPC differences between Scenarios 3 and 4. The results of these tests indicate that the choice of ESS technology for the MG is not immediately apparent.The most suitable ESS technology will depend on the specific MCS and reliability requirements of the MG.F I G U R E 22 MCMG capacity shortage under multiple scenarios while MCS is assumed to be 9%.(A) Scenario 1, (B) Scenario 2, (C) Scenario 3, and (D) Scenario 4. MCMG, multicarrier microgrid; MCS, maximum capacity shortage.