Analyzing the Impact of Renewable Energy Incentives and Parameter Uncertainties on Financial Feasibility of a Campus Microgrid

: The popularity of microgrids is increasing considerably because of their environmental and technical advantages. However, the major challenge in microgrid integration is its ﬁnancial feasibility due to high capital costs. To address this obstacle, renewable energy incentive programs, which are the motivation of this study, have been proposed in many countries. This paper provides a comprehensive evaluation of the technical and ﬁnancial feasibility of a campus microgrid based on a techno-economic analysis using the Microgrid Decision Support Tool, which was implemented to support decision-making in the context of microgrid project investment. A method for microgrid design aiming to maximize system proﬁtability is presented. The optimal microgrid conﬁguration is selected depending on ﬁnancial indices of the project, which directly address the returns on an investment. Most importantly, this analysis captures all the beneﬁts of ﬁnancial incentives for microgrid projects in California, U.S., which presents a key difference between the California market and other markets. The impact of incentives and uncertain ﬁnancial parameters on the project investment is veriﬁed by sensitivity analysis. The outcomes show that the optimal conﬁguration generates signiﬁcant electricity savings, and the incentives strongly determine the ﬁnancial feasibility and the optimal design of a microgrid.


Introduction
In recent years, electric power system design has witnessed a paradigm shift from the traditional centralized grid system toward decentralized and independent systems called microgrids, which can operate in either the grid-connected or islanded mode to supply demand targets. Researchers have increasingly been focusing on microgrids because they offer obvious advantages, such as high penetration of renewable energy resources (RESs), improved reliability due to the autonomous operation of distributed energy resources (DERs), and significant reductions in greenhouse gas emissions as well as reductions in a power system's operating costs. Specifically, the optimal design of a microgrid considering economic and dynamic performance analysis was presented in reference [1], which contains photovoltaics (PV), wind turbines, a battery energy storage system (BESS), and two diesel generators. The study indicated important benefits by microgrid adoption such as minimization of levelized cost of energy (LCOE), maximization of renewable energy penetration and significant

Performance Model
The performance model simulates one-year power generation and consumption for each microgrid configuration in a search space which contains all of the possible capacities and combinations of DERs defined by the user. The inputs for the performance model include 8760-h time-series data of electricity load demands and weather resources, such as solar irradiance and wind speed, the technical specifications and cost parameters of DERs and a user-defined economic dispatch algorithm. The output is a set of time-series data which details hourly energy production and consumption within the system. The mathematical modeling of DERs is the primary determinant of the accuracy of this model. MDSTool employs time-series behavioral models used for long-term performance prediction and evaluation. Currently, the technologies modeled in MDSTool consist of PV, wind turbines, geothermal energy, biomass, internal combustion engines, FCs and energy storage. As the microgrid used as a case study in this paper contains a PV, a BESS, and an FC, their mathematical models are presented here.

PV Model
PV modules are rated on the basis of the power generated under the standard testing condition (STC) of 1 kW/m 2 of sunlight and a PV cell temperature of 25 °C. Various performance models for PV modules were developed [24,25]. The power produced by PV modules can be calculated as a function of solar irradiance, as given in Equation (1).

BESS Model
MDSTool uses the battery model developed by Manwell and McGowan [26], which is based on the concept of electrochemical kinetics, and can determine the charge and discharge power from a battery at each time step. The maximum amount of power that a battery can discharge and charge over time step Δt [h] is given by Equations (2) and (3), respectively.

Performance Model
The performance model simulates one-year power generation and consumption for each microgrid configuration in a search space which contains all of the possible capacities and combinations of DERs defined by the user. The inputs for the performance model include 8760-h time-series data of electricity load demands and weather resources, such as solar irradiance and wind speed, the technical specifications and cost parameters of DERs and a user-defined economic dispatch algorithm. The output is a set of time-series data which details hourly energy production and consumption within the system. The mathematical modeling of DERs is the primary determinant of the accuracy of this model. MDSTool employs time-series behavioral models used for long-term performance prediction and evaluation. Currently, the technologies modeled in MDSTool consist of PV, wind turbines, geothermal energy, biomass, internal combustion engines, FCs and energy storage. As the microgrid used as a case study in this paper contains a PV, a BESS, and an FC, their mathematical models are presented here.

PV Model
PV modules are rated on the basis of the power generated under the standard testing condition (STC) of 1 kW/m 2 of sunlight and a PV cell temperature of 25 • C. Various performance models for PV modules were developed [24,25]. The power produced by PV modules can be calculated as a function of solar irradiance, as given in Equation (1).

BESS Model
MDSTool uses the battery model developed by Manwell and McGowan [26], which is based on the concept of electrochemical kinetics, and can determine the charge and discharge power from a Energies 2018, 11,2446 5 of 24 battery at each time step. The maximum amount of power that a battery can discharge and charge over time step ∆t [h] is given by Equations (2) and (3), respectively.
A power electronic converter is modeled as a black box and the modeling is undertaken using an efficiency, which reflects the power loss between the input and output.

FC Model
FC is considered to be generator with generation cost model using fuel cost curve, in which the fuel consumption as a function of the electrical output power is assumed to be linear as in reference [27]:

DER Dispatch Model
To perform the system operation optimization, a dispatch algorithm for DERs must be specified. This dispatch algorithm involves the problem of unit commitment and economic dispatch of DERs. At each time step, the algorithm decides which components will operate and at what power level to efficiently supply the load demand at a minimum operating cost while satisfying operational constraints. The objective function is described by Equation (5) followed by its constraints. minC total,t = C gen,t + C dch,t + C gbuy,t − C gsell,t − C dres,t − C anc,t subject to the following constraints: P load,t = P G,t + P S,t + P gbuy,t − P gsell,t SoC min ≤ SoC t ≤ SoC max (8) Equation (6) describes the energy balance constraint between energy demand and supply at each time step. The set of Equation (7) shows the inequality constraints controlling the output power of the energy sources, as well as the purchase from and sale to the grid to satisfy the maximum and minimum thresholds. Equation (8) indicates that the SoC of storage units must be maintained between the maximum and minimum values. The optimization of system operation is performed at each time step, and this is repeated 8760 times to complete the one-year operation of a configuration. Subsequently, the performance model decides whether a configuration is technically feasible based on reliability constraints. MDSTool utilizes the two most widely used reliability metrics: Loss-of-load probability (LOLP) and loss-of-power supply probability (LPSP) [28]. As a result, the energy generation and consumption profiles of feasible configurations are saved and used as the inputs for the economic model, which calculates all cash flows for financial feasibility and optimal sizing analyses.

Economic Model
MDSTool uses cash flow analysis, which accounts for all expenses (cash outflows) and benefits (cash inflows). The pre-tax cash flow calculates the total installation and operating costs, and all project revenues, while the after-tax cash flow includes all debt payments and income tax considering tax benefits from debt interest payments, depreciation and tax credits such as ITC and production tax credit (PTC) [22].
The pre-tax cash outflows and inflows are calculated using Equations (9) and (10), respectively. The only pre-tax cash outflow in year zero is the project equity, while the cash outflows in subsequent years account for all costs incurred over the project's lifetime. The pre-tax cash inflow in year zero is equal to any IBI. In subsequent years, the pre-tax cash inflows are all project revenues.
where C annual = C o&m + C gbuy The total pre-tax cost in each year is the net difference between pre-tax cash inflows and cash outflows. The total pre-tax cash flow is calculated by adding the benefits from energy savings to the pre-tax cost.
F PTCF,n = F PTC,n + C saving .(1 + r) n (12) In the after-tax basis, the total after-tax cost is determined by subtracting the total debt payment and income tax from F PTC , while the total after-tax cash flow is determined by adding after-tax savings to F ATC . To capture tax benefits for tax-paying investors, the calculation of income tax accounts for tax deductions from debt interest payments, depreciation, and tax credits for DERs [22].
F ATCF,n = F ATC,n + δ · C saving · (1 + r) n (14) The total costs and cash flows can be used to calculate different financial indices used for optimal selection of DERs and financial feasibility analysis. MDSTool provides standard financial metrics such as NPV, lifecycle cost (LCC), benefit-cost ratio (BCR), simple payback period (SPB), internal rate of return (IRR), and annual lifecycle savings (ALCS). Hereby, MDSTool provides both pre-tax and after-tax cash flow analysis. The choice of pre-tax or after-tax analysis depends on the ownership of a microgrid project. Pre-tax financial analysis is appropriate for non-tax-paying investors such as governmental agencies, while after-tax analysis is crucial for private entities investing in microgrids because taxes and tax credits are real costs which directly affect cost and revenue streams of a commercial investment.

Microgrid Design Using MDSTool
The purpose of this study is to provide an optimal microgrid design methodology for commercial ownership based on financial feasibility analysis, which can capture all of the benefits from renewable Energies 2018, 11, 2446 7 of 24 energy incentives and tax reduction programs and evaluates their impact on the investment. For this reason, this method uses after-tax basis to compare alternative configurations.
With the objective of addressing all benefits and returns by a microgrid deployment, the metrics used in this study for financial feasibility evaluation are NPV, SPB, LCC, and BCR. NPV is computed by subtracting the present values of cash outflows from the present values of cash inflows over the lifetime of the investment, as shown in Equation (16). It is a direct measurement that determines to what extent an investment will be profitable. In the context of investment decision-making, it has been shown that NPV leads to better decisions than other indices [29]. For this reason, NPV is selected as the primary ranking criteria to find the optimal configuration, which is the most profitable case (i.e., that with the highest NPV). In the case where there are alternative configurations which produce the same benefits and returns, the other secondary metrics will need to be evaluated. LCC is a measurement of all costs incurred during a project's lifetime. BCR is defined as the ratio of the present value to the initial installation cost as calculated in Equation (18).
The SPB of a project is determined as the number of years it takes before the cumulative forecasted cash flow equals or exceeds the initial investment. It can be expressed as the first point in the project's lifetime at which: Figure 2 presents the microgrid design process using MDSTool, which is divided into two main sections: (1) Problem formulation and (2) simulation and analysis. In the first section, definitions of the design objective and the DER search space are decisive steps. The objective of a microgrid design must be clearly defined because it will decide which evaluation parameters will be used to search for the optimal configuration. In this case study, the design objective function is to maximize the economic benefits, therefore, NPV is used for the optimal DER sizing analysis. The search space in which the optimal DER capacity is located includes all of the possible capacities of the selected DERs considering associated technical and financial constraints. For example, the rated power of PV systems is typically constrained by the available installation space, especially for campus microgrids where PV modules are often installed on the rooftops of buildings [16]. MDSTool will perform the system operation optimization to find the optimal design for further analysis.
Energies 2018, 11, x FOR PEER REVIEW 7 of 24 from renewable energy incentives and tax reduction programs and evaluates their impact on the investment. For this reason, this method uses after-tax basis to compare alternative configurations.
With the objective of addressing all benefits and returns by a microgrid deployment, the metrics used in this study for financial feasibility evaluation are NPV, SPB, LCC, and BCR. NPV is computed by subtracting the present values of cash outflows from the present values of cash inflows over the lifetime of the investment, as shown in Equation (16). It is a direct measurement that determines to what extent an investment will be profitable. In the context of investment decision-making, it has been shown that NPV leads to better decisions than other indices [29]. For this reason, NPV is selected as the primary ranking criteria to find the optimal configuration, which is the most profitable case (i.e., that with the highest NPV). In the case where there are alternative configurations which produce the same benefits and returns, the other secondary metrics will need to be evaluated. LCC is a measurement of all costs incurred during a project's lifetime. BCR is defined as the ratio of the present value to the initial installation cost as calculated in Equation (18).
The SPB of a project is determined as the number of years it takes before the cumulative forecasted cash flow equals or exceeds the initial investment. It can be expressed as the first point in the project's lifetime at which: Figure 2 presents the microgrid design process using MDSTool, which is divided into two main sections: (1) Problem formulation and (2) simulation and analysis. In the first section, definitions of the design objective and the DER search space are decisive steps. The objective of a microgrid design must be clearly defined because it will decide which evaluation parameters will be used to search for the optimal configuration. In this case study, the design objective function is to maximize the economic benefits, therefore, NPV is used for the optimal DER sizing analysis. The search space in which the optimal DER capacity is located includes all of the possible capacities of the selected DERs considering associated technical and financial constraints. For example, the rated power of PV systems is typically constrained by the available installation space, especially for campus microgrids where PV modules are often installed on the rooftops of buildings [16]. MDSTool will perform the system operation optimization to find the optimal design for further analysis.

Case Study of a Campus Microgrid in California, U.S.
The presented design method using MDSTool was applied to determine the optimal design for a university campus microgrid in California, U.S. The microgrid configuration, shown in Figure 3, comprises of a PV, an ESS using lithium-ion batteries and an FC. The microgrid is connected to the San Diego Gas and Electric (SDG&E) utility, and it is designed to power a high school, a university dormitory and a research building. This section analyzes all the data required to run this case study in MDSTool.

Case Study of a Campus Microgrid in California, U.S.
The presented design method using MDSTool was applied to determine the optimal design for a university campus microgrid in California, U.S. The microgrid configuration, shown in Figure 3, comprises of a PV, an ESS using lithium-ion batteries and an FC. The microgrid is connected to the San Diego Gas and Electric (SDG&E) utility, and it is designed to power a high school, a university dormitory and a research building. This section analyzes all the data required to run this case study in MDSTool.

Electric Tariff
The applied electric tariff is AL-TOU of SDG&E utility, which is applicable to all metered nonresidential customers, whose monthly maximum demand equals, exceeds, or is expected to equal or exceed 20 kW. This rate structure employs time-of-use (TOU) energy charges in $/kWh and a monthly demand charge of 24.51 $/kW [30]. The AL-TOU tariffs for November-April and May-October, and for weekdays/weekends are shown in Figure 4. The numbers in blue indicate the rates presented in Table 1, with the highest rate of 0.11703 $/kWh and the lowest rate of 0.07745 $/kWh [30]. There are no peak and off-peak hours during the weekend. Notably, the first six hours and the last hour of each weekday are the periods with the lowest power prices. This influences the selection of an appropriate dispatch algorithm for the ESS to benefit from the AL-TOU tariff. In addition, the very high monthly demand charge of 24.51 $/kWh highlights the potential profit that can be generated by reducing demand charge via the microgrid.

Electric Tariff
The applied electric tariff is AL-TOU of SDG&E utility, which is applicable to all metered non-residential customers, whose monthly maximum demand equals, exceeds, or is expected to equal or exceed 20 kW. This rate structure employs time-of-use (TOU) energy charges in $/kWh and a monthly demand charge of 24.51 $/kW [30]. The AL-TOU tariffs for November-April and May-October, and for weekdays/weekends are shown in Figure 4. The numbers in blue indicate the rates presented in Table 1, with the highest rate of 0.11703 $/kWh and the lowest rate of 0.07745 $/kWh [30]. There are no peak and off-peak hours during the weekend. Notably, the first six hours and the last hour of each weekday are the periods with the lowest power prices. This influences the selection of an appropriate dispatch algorithm for the ESS to benefit from the AL-TOU tariff. In addition, the very high monthly demand charge of 24.51 $/kWh highlights the potential profit that can be generated by reducing demand charge via the microgrid.

Case Study of a Campus Microgrid in California, U.S.
The presented design method using MDSTool was applied to determine the optimal design for a university campus microgrid in California, U.S. The microgrid configuration, shown in Figure 3, comprises of a PV, an ESS using lithium-ion batteries and an FC. The microgrid is connected to the San Diego Gas and Electric (SDG&E) utility, and it is designed to power a high school, a university dormitory and a research building. This section analyzes all the data required to run this case study in MDSTool.

Electric Tariff
The applied electric tariff is AL-TOU of SDG&E utility, which is applicable to all metered nonresidential customers, whose monthly maximum demand equals, exceeds, or is expected to equal or exceed 20 kW. This rate structure employs time-of-use (TOU) energy charges in $/kWh and a monthly demand charge of 24.51 $/kW [30]. The AL-TOU tariffs for November-April and May-October, and for weekdays/weekends are shown in Figure 4. The numbers in blue indicate the rates presented in Table 1, with the highest rate of 0.11703 $/kWh and the lowest rate of 0.07745 $/kWh [30]. There are no peak and off-peak hours during the weekend. Notably, the first six hours and the last hour of each weekday are the periods with the lowest power prices. This influences the selection of an appropriate dispatch algorithm for the ESS to benefit from the AL-TOU tariff. In addition, the very high monthly demand charge of 24.51 $/kWh highlights the potential profit that can be generated by reducing demand charge via the microgrid.

Electricity Load Demand
In this study, the economic benefit of DSM is evaluated because it would substantially improve the economic security of the microgrid. To achieve this, two load profiles were created, including the base load and the controlled load. The system base load is assumed to be the existing load profile, whereas the controlled load is the expected load profile after applying load control, which results in the overall load reduction and load shifting.

Base Load Data Generation
The electricity demand data for the three institutions attached to the microgrid-Preuss school, a dormitory for graduate students, and a research building called California Institute for Telecommunications and Information Technology, all located in the University of California, San Diego-were measured for three months (May, June, and July 2017). Based on these onsite measurement data, load profiles for an entire year, with a resolution of one hour, were generated for each type of load with the following assumptions: The load shape of each building is conserved by scaling the measured data; the total load reaches a maximum in March and September and a minimum in January and December; and the maximum, average, and minimum loads of the overall load profile are 1011 kW, 572 kW, and 247 kW, respectively. Table 2 summarizes the load parameters of each load type. Notably, the maximum and minimum loads of the total demand are different from the summation of the individual values because the maximum and minimum values of the three loads can occur at different times on any given day. Figures 5 and 6 show the typical daily load profiles and the total energy seasonal variations, respectively, of the electricity demands of the three loads. Figure 7 presents the total hourly electricity demand for a complete year.

Electricity Load Demand
In this study, the economic benefit of DSM is evaluated because it would substantially improve the economic security of the microgrid. To achieve this, two load profiles were created, including the base load and the controlled load. The system base load is assumed to be the existing load profile, whereas the controlled load is the expected load profile after applying load control, which results in the overall load reduction and load shifting.

Base Load Data Generation
The electricity demand data for the three institutions attached to the microgrid-Preuss school, a dormitory for graduate students, and a research building called California Institute for Telecommunications and Information Technology, all located in the University of California, San Diego-were measured for three months (May, June, and July 2017). Based on these onsite measurement data, load profiles for an entire year, with a resolution of one hour ,were generated for each type of load with the following assumptions: The load shape of each building is conserved by scaling the measured data; the total load reaches a maximum in March and September and a minimum in January and December; and the maximum, average, and minimum loads of the overall load profile are 1011 kW, 572 kW, and 247 kW, respectively. Table 2 summarizes the load parameters of each load type. Notably, the maximum and minimum loads of the total demand are different from the summation of the individual values because the maximum and minimum values of the three loads can occur at different times on any given day. Figures 5 and 6 show the typical daily load profiles and the total energy seasonal variations, respectively, of the electricity demands of the three loads. Figure 7 presents the total hourly electricity demand for a complete year.    One of the important parameters of the total load demand is its high load factor of 0.57. This means that the difference between the average and peak load is relatively low. The high load factor significantly influences the selection of the appropriate dispatch strategy and optimal sizing of the DER units, because a substantial proportion of the electricity savings generated by the microgrid deployment comes from energy charges rather than demand charges.

Controlled Load with DSM
The concept of DSM is executed by installing smart sensors, lighting control systems (LCS), electric heat pumps (EHP), and building energy management systems (BEMS). Details of DSM operation are not presented in this paper, but the associated latter assumptions and installation costs are based on the measurement data obtained from our previous microgrid project, the Internet-of-Things (IoT)-based campus microgrid at Seoul National University (SNU), South Korea, with a peak load of 912 kW [22]. DSM contributes to overall demand reduction and demand shifting, which shifts a portion of the load from periods with peak electricity prices to those with lower prices. Based on the performance of the SNU microgrid, the assumptions of load control with DSM for this study were established. The total demand was reduced by 8.9%, and 5% of the load from 11:00 a.m. to 05:00 p.m. was uniformly shifted to the period from 00:00 a.m. to 06:00 a.m. every weekday because this period has the lowest electricity prices. The demand over the weekend was not shifted. The load profile created from the base load data using these assumptions is defined as the controlled load.  One of the important parameters of the total load demand is its high load factor of 0.57. This means that the difference between the average and peak load is relatively low. The high load factor significantly influences the selection of the appropriate dispatch strategy and optimal sizing of the DER units, because a substantial proportion of the electricity savings generated by the microgrid deployment comes from energy charges rather than demand charges.

Controlled Load with DSM
The concept of DSM is executed by installing smart sensors, lighting control systems (LCS), electric heat pumps (EHP), and building energy management systems (BEMS). Details of DSM operation are not presented in this paper, but the associated latter assumptions and installation costs are based on the measurement data obtained from our previous microgrid project, the Internet-of-Things (IoT)-based campus microgrid at Seoul National University (SNU), South Korea, with a peak load of 912 kW [22]. DSM contributes to overall demand reduction and demand shifting, which shifts a portion of the load from periods with peak electricity prices to those with lower prices. Based on the performance of the SNU microgrid, the assumptions of load control with DSM for this study were established. The total demand was reduced by 8.9%, and 5% of the load from 11:00 a.m. to 05:00 p.m. was uniformly shifted to the period from 00:00 a.m. to 06:00 a.m. every weekday because this period has the lowest electricity prices. The demand over the weekend was not shifted. The load profile created from the base load data using these assumptions is defined as the controlled load. One of the important parameters of the total load demand is its high load factor of 0.57. This means that the difference between the average and peak load is relatively low. The high load factor significantly influences the selection of the appropriate dispatch strategy and optimal sizing of the DER units, because a substantial proportion of the electricity savings generated by the microgrid deployment comes from energy charges rather than demand charges.

Controlled Load with DSM
The concept of DSM is executed by installing smart sensors, lighting control systems (LCS), electric heat pumps (EHP), and building energy management systems (BEMS). Details of DSM operation are not presented in this paper, but the associated latter assumptions and installation costs are based on the measurement data obtained from our previous microgrid project, the Internet-of-Things (IoT)-based campus microgrid at Seoul National University (SNU), South Korea, with a peak load of 912 kW [22]. DSM contributes to overall demand reduction and demand shifting, which shifts a portion of the load from periods with peak electricity prices to those with lower prices. Based on the performance of the SNU microgrid, the assumptions of load control with DSM for this study were established. The total demand was reduced by 8.9%, and 5% of the load from 11:00 a.m. to 05:00 p.m. was uniformly shifted to the period from 00:00 a.m. to 06:00 a.m. every weekday because this period has the lowest electricity prices. The demand over the weekend was not shifted. The load profile created from the base load data using these assumptions is defined as the controlled load.

Weather Resources
The data of the average Global Horizontal Irradiance (GHI), and temperature at the University of California, San Diego, was downloaded from the NASA Surface Meteorology and Solar Energy website [31]. Based on this data, 8760-h time-series data was generated to be used for simulations in MDSTool. Figure 8 presents the GHI over a complete year by hour.

Weather Resources
The data of the average Global Horizontal Irradiance (GHI), and temperature at the University of California, San Diego, was downloaded from the NASA Surface Meteorology and Solar Energy website [31]. Based on this data, 8760-h time-series data was generated to be used for simulations in MDSTool. Figure 8 presents the GHI over a complete year by hour.

Renewable Energy Incentives
To evaluate the financial feasibility of the microgrid project, all of the applicable incentives and tax credits were analyzed in this study. California's Renewables Portfolio Standard requires Californian electric utilities to have 33% of their retail sales derived from eligible renewable energy resources by 31 December 2020, 40% by 31 December 2024, 45% by 31 December 2017, and 50% by 31 December 2030 and all subsequent years [32]. This standard is one of the major factors driving the establishment of various state incentives for renewable energy projects, which are the main motivation of this paper.
The campus microgrid is eligible to receive federal tax credit and PTE on its PV system, and IBI under California self-generation incentive program (SGIP) for its ESS and FC. Other incentives considered include NEM and MACRS. Table 3 summarizes the incentives available in San Diego, California. Notably, the feed-in tariff, which is renewable market adjusting tariff (ReMAT), is applicable in California. However, according to the ReMAT policy, any customer who sells power to the utility under this tariff cannot participate in other state incentive programs [33]. In this case, if the microgrid project were to be enrolled under ReMAT, it would not be eligible for the SGIP, NEM, and PTE programs. For this reason, the incentive programs described in Table 3 were analyzed and included in the project financial analysis.

Renewable Energy Incentives
To evaluate the financial feasibility of the microgrid project, all of the applicable incentives and tax credits were analyzed in this study. California's Renewables Portfolio Standard requires Californian electric utilities to have 33% of their retail sales derived from eligible renewable energy resources by 31 December 2020, 40% by 31 December 2024, 45% by 31 December 2017, and 50% by 31 December 2030 and all subsequent years [32]. This standard is one of the major factors driving the establishment of various state incentives for renewable energy projects, which are the main motivation of this paper.
The campus microgrid is eligible to receive federal tax credit and PTE on its PV system, and IBI under California self-generation incentive program (SGIP) for its ESS and FC. Other incentives considered include NEM and MACRS. Table 3 summarizes the incentives available in San Diego, California. Notably, the feed-in tariff, which is renewable market adjusting tariff (ReMAT), is applicable in California. However, according to the ReMAT policy, any customer who sells power to the utility under this tariff cannot participate in other state incentive programs [33]. In this case, if the microgrid project were to be enrolled under ReMAT, it would not be eligible for the SGIP, NEM, and PTE programs. For this reason, the incentive programs described in Table 3 were analyzed and included in the project financial analysis. Energy tax credits for renewables substantially enhance after-tax cash flow and promote an investment. An energy ITC is calculated as an immediate reduction in income taxes equal to a percentage of the installed cost of a new investment. The ITCs for solar and FC technologies are summarized in Table 4, wherein the specified date is the expiration date in each year [34]. Notably, the expiration date for solar technologies is based on when the project construction starts. The ITC for PV systems remains constant at 30% until December 2019, after which it will progressively reduce to 10% from 2022 onward, with no maximum credit, whereas the ITC for FCs ended in 2016 [34]. The PV system components eligible for the ITC include the ESS and the power conditioning system.

California Self-Generation Incentive Program
SGIP is available to the customers of four utilities in California: SDG&E, Pacific Gas and Electric, Southern California Edison, and Southern California Gas (SoCal Gas). The objective of this incentive program is to encourage the installation of new and sustainable technologies, thereby contributing to a reduction of greenhouse gases, demands, and customer electricity purchases. SGIP provides financial incentives, which are in the form of IBIs, to customers of the aforementioned utilities who generate electricity with wind turbines, FCs using renewable and non-renewable fuels, various forms of combined heat and power, and advanced energy storage. The incentive rates of SGIP for FCs and ESSs are detailed in Tables 5 and 6, respectively [35].
In a year, the total SGIP fund is divided into incentive steps that differ in their incentive rates. The availability of funding steps depends on the number of projects that have applied for this incentive. In the situation where there are many projects, if the funding of a step ends, the next step will be opened with lower incentive rates. For FC technology, the total generation incentive fund is divided into three steps, and the incentive rates decrease by 0.1 $/W between each step. For energy storage, the total incentive fund is divided into five steps. The reduction in incentive rates based on energy storage capacity and generation capacity are shown in Tables 7 and 8, respectively [35]. Energy storage incentives are available for up to the storage capacity of 6 MWh, and FC incentives are paid for up to 3 MW both with tiered incentive rates. The SGIP incentive rates of the last funding step, which are the lowest values, were included in the economic model of MDSTool so that the results were more conservative. Variations in the SGIP rates were evaluated in the sensitivity analysis. Table 5. Generation incentive rates per Watt for fuel cell (FC) in SGIP.

Net Energy Metering
Residential and commercial customers of SDG&E whose electricity generating systems supply their onsite electricity demand in full or partly using solar, wind, biomass, geothermal, or small hydro are eligible for the NEM program of California. NEM provides customers full retail rate credits for any excess power generated by their system that enters the grid [36]. In this case study, the abovementioned AL-TOU tariff is applied to NEM. Participation in NEM does not restrict a customer's eligibility for any other rebate, incentive, or credit provided by an electric utility [36].

Modified Accelerated Cost-Recovery System
For tax purposes, depreciation is a means of recovering, through an income tax deduction, the cost of property used in a trade or business, or of property held for the production of income. Under modified accelerated cost-recovery system (MACRS), the project may recover investments on a certain property through depreciation deductions. In this case study, the FC and the PV system, including the ESS, are eligible for five-year MACRS [37]. The depreciation rates for recovery periods from the first year to the sixth year are 20.00%, 32.00%, 19.20%, 11.52%, 11.52%, and 5.76%, respectively [37].

Property Tax Exclusion for Solar Energy System
Section 73 of the California Revenue and Taxation Code allows property tax exclusion (PTE) for active solar energy systems installed between 1 January 1999 and 31 December 2024 [38]. An active solar energy system is defined as a system that uses solar devices, which are thermally isolated from the living space or any other area where the energy is used, to facilitate the collection, storage, or distribution of solar energy. Such a system contains storage devices, power conditioning equipment, transfer equipment, and components related to the functioning of these items. The systems that qualify for this incentive include solar space conditioning systems, solar water heating systems, photovoltaic systems, solar thermal electric systems and solar mechanical energy. Solar swimming pool heaters, hot tub heaters, passive energy systems and wind energy systems are not eligible for this incentive. An active solar energy system does not contain any auxiliary equipment that uses a source other than solar energy to provide usable energy. In this case study, the PV system, including the ESS and converter, is eligible for 100% exclusion [38].

Component Parameters and Financial Parameters
The input data of component parameters and the project financial parameters are detailed in Tables 9 and 10, respectively. The capital costs of DERs vary widely depending on many factors such as the manufacturer, rated capacity, market regulations, and technical features. The project financial parameters include a high level of uncertainty as they depend on the local economy. For the present study, the nominal discount rate was assumed to be 8% [16]. The expected inflation rate of 2.15% was the average value of the historical U.S. consumer price index over the previous 20 years. For commercial ownership, the project was assumed to be 100% financed by a long-term loan.

Optimal Configuration
As described above in the microgrid architecture, the campus microgrid may comprise of a PV, an FC, or an ESS. As such, this study considered two options: "PV + ESS" and "PV + FC + ESS". The optimal design for each of these options was determined and compared to select the optimal microgrid configuration. The energy production and consumption of the two microgrid configurations were compared to that of the base case, which is assumed to be the existing system wherein the entire load demand is supplied by the SDG&E utility. In addition, to evaluate the economics of load control, the base load was used to simulate the base case, while the microgrid cases were simulated using the controlled load. To simulate the case study in MDSTool, in addition to the input data presented above, a search space must be defined to locate the optimal configuration. The search space of PV ranged from 0 to 300 kW, that of ESS ranged from 0 to 1000 kWh and that of its converter ranged from 0 to 500 kW, all in steps of 25 kW. The rated power of FC was constrained by the total capital cost. The energy outputs of the PV + ESS and PV + FC + ESS microgrid configurations are illustrated in Figures 9 and 10, respectively. In both cases, the ESS was operated by a peak-shaving dispatch strategy, which discharges the ESS during peak-load periods and charges the ESS in periods with low electricity rates to reduce monthly demand charges. Specifically, with the AL-TOU tariff, the ESS charges during the first six hours of a day and discharges to flatten grid purchases as shown in the two figures. The period of high power demand in a day often coincides with the on-peak utility billing period. In the PV + FC + ESS configuration, the FC operates at full power to cover a proportion of the total load, while the rest of the load is supplied by the PV, utility, or ESS discharge. This operation of FC resulted in significant energy savings.
wherein the entire load demand is supplied by the SDG&E utility. In addition, to evaluate the economics of load control, the base load was used to simulate the base case, while the microgrid cases were simulated using the controlled load. To simulate the case study in MDSTool, in addition to the input data presented above, a search space must be defined to locate the optimal configuration. The search space of PV ranged from 0 to 300 kW, that of ESS ranged from 0 to 1000 kWh and that of its converter ranged from 0 to 500 kW, all in steps of 25 kW. The rated power of FC was constrained by the total capital cost.
The energy outputs of the PV + ESS and PV + FC + ESS microgrid configurations are illustrated in Figures 9 and 10, respectively. In both cases, the ESS was operated by a peak-shaving dispatch strategy, which discharges the ESS during peak-load periods and charges the ESS in periods with low electricity rates to reduce monthly demand charges. Specifically, with the AL-TOU tariff, the ESS charges during the first six hours of a day and discharges to flatten grid purchases as shown in the two figures. The period of high power demand in a day often coincides with the on-peak utility billing period. In the PV + FC + ESS configuration, the FC operates at full power to cover a proportion of the total load, while the rest of the load is supplied by the PV, utility, or ESS discharge. This operation of FC resulted in significant energy savings.   wherein the entire load demand is supplied by the SDG&E utility. In addition, to evaluate the economics of load control, the base load was used to simulate the base case, while the microgrid cases were simulated using the controlled load. To simulate the case study in MDSTool, in addition to the input data presented above, a search space must be defined to locate the optimal configuration. The search space of PV ranged from 0 to 300 kW, that of ESS ranged from 0 to 1000 kWh and that of its converter ranged from 0 to 500 kW, all in steps of 25 kW. The rated power of FC was constrained by the total capital cost.
The energy outputs of the PV + ESS and PV + FC + ESS microgrid configurations are illustrated in Figures 9 and 10, respectively. In both cases, the ESS was operated by a peak-shaving dispatch strategy, which discharges the ESS during peak-load periods and charges the ESS in periods with low electricity rates to reduce monthly demand charges. Specifically, with the AL-TOU tariff, the ESS charges during the first six hours of a day and discharges to flatten grid purchases as shown in the two figures. The period of high power demand in a day often coincides with the on-peak utility billing period. In the PV + FC + ESS configuration, the FC operates at full power to cover a proportion of the total load, while the rest of the load is supplied by the PV, utility, or ESS discharge. This operation of FC resulted in significant energy savings.    Table 11 details the breakdown of energy production and electricity charges of the two optimal microgrid configurations compared to the base case. In the base case, the total annual energy charge is $489,584.94, whereas the annual demand charge is $250,996.72, which constitutes more than half of the energy charge. The noticeably high annual demand charge is partly owing to the high monthly demand charge levied by the AL-TOU tariff, which is 24.51 $/kW. With the optimal PV + ESS configuration supplying the controlled load, the annual energy charge and demand charge were reduced to $378,943.16 and $180,946.86, respectively, resulting in the annual electricity saving of $180,691.64. In the PV + FC + ESS configuration, the annual energy and demand charges were reduced further because a significant proportion of the grid purchase was replaced by the FC generation. Indeed, the annual energy charge and demand charge were greatly curtailed to $242,823.93 and $133,320.26, respectively, leading to annual electricity savings of up to $364,437.47 (49.21%). However, an amount of $121,413.60 needed to be paid for annual fuel cost in this case. Table 12 compares the financial feasibility of the two optimal microgrid cases, and a breakdown of their associated costs and benefits is illustrated in Figure 11. The capital cost of PV + ESS configuration was $1,275,750.00, and this cost was used as a constraint on the investment size when determining the optimal PV + FC + ESS configuration in order to establish which configuration had the higher profitability. The PV + FC + ESS case received more grant incentives because both the FC and the ESS are eligible for the SGIP. With approximately the same capital costs, the PV + FC + ESS configuration produced an NPV of $1,594,000.00, which is $528,800.00 higher than that of the PV+ESS configuration. This leads to a notable difference in the resulting BCRs. The PV + FC + ESS configuration has a BCR of 2.25, while the other has a BCR of just 1.84. Both cases have an SPB of around 5 y. Based on this comparison, it was concluded that the financial performance of PV + FC + ESS configuration is superior to that of the PV + ESS configuration. As a result, the PV + FC + ESS microgrid with the FC of 220 kW, PV of 50 kW, and ESS of 300 kWh was selected as the optimal microgrid configuration for the sensitivity analysis.

Incentives
As described in the SGIP section, the incentive rates for the ESS and FC vary according to predetermined incentive steps, and this variation is considered in this section. The SGIP rate for large storage units claiming ITC declines from 0.36 $/Wh to 0.22 $/Wh, and that for the FC decreases by 0.1 $/W between incentive steps. As presented in Table 4 with regards to the ITC for PV systems, the incentive rate will be adjusted in the future. PV projects can obtain an ITC of 30% if construction starts before 12 December 2019, after which the rate will decrease by 4% each year until 12 December 2021 and remain at 10% thereafter. In addition to these considerations, the scenarios without each of the ITC, SGIP, or PTE, and a scenario without all the three incentives were simulated to examine their impact on the financial feasibility of the optimal configuration and the sizing of DERs. Table 13 presents the impact of variation in incentive levels and their absence on the financial indices of the optimal configuration. It is apparent that the higher the incentive rates, the more profitable the project becomes, with a higher NPV and a smaller SPB. For the ITC, the project construction should start soon to obtain a higher ITC value. When the ITC declines to 10%, the NPV will decrease by $76,800.00. The SGIP strongly enhances the financial feasibility of the microgrid. In the situation where both the FC and ESS can receive the highest SGIP rates, the optimal configuration produces an NPV of up to $1,721,600.00, with the SPB decreasing to 4.03 y. Of the three incentives, the PTE slightly affects the project's NPV when it is not applied. In contrast, the absence of the SGIP has dominant influences. Most importantly, if all of the three incentives are removed, the microgrid's SPB will be more than 7 y and its NPV will decrease by nearly half, which takes the BCR down to just 1.54.

Incentives
As described in the SGIP section, the incentive rates for the ESS and FC vary according to predetermined incentive steps, and this variation is considered in this section. The SGIP rate for large storage units claiming ITC declines from 0.36 $/Wh to 0.22 $/Wh, and that for the FC decreases by 0.1 $/W between incentive steps. As presented in Table 4 with regards to the ITC for PV systems, the incentive rate will be adjusted in the future. PV projects can obtain an ITC of 30% if construction starts before 12 December 2019, after which the rate will decrease by 4% each year until 12 December 2021 and remain at 10% thereafter. In addition to these considerations, the scenarios without each of the ITC, SGIP, or PTE, and a scenario without all the three incentives were simulated to examine their impact on the financial feasibility of the optimal configuration and the sizing of DERs. Table 13 presents the impact of variation in incentive levels and their absence on the financial indices of the optimal configuration. It is apparent that the higher the incentive rates, the more profitable the project becomes, with a higher NPV and a smaller SPB. For the ITC, the project construction should start soon to obtain a higher ITC value. When the ITC declines to 10%, the NPV will decrease by $76,800.00. The SGIP strongly enhances the financial feasibility of the microgrid. In the situation where both the FC and ESS can receive the highest SGIP rates, the optimal configuration produces an NPV of up to $1,721,600.00, with the SPB decreasing to 4.03 y. Of the three incentives, the PTE slightly affects the project's NPV when it is not applied. In contrast, the absence of the SGIP has dominant influences. Most importantly, if all of the three incentives are removed, the microgrid's SPB will be more than 7 y and its NPV will decrease by nearly half, which takes the BCR down to just 1.54. Table 14 presents the impact of excluding incentives on the optimal sizing of the DERs. It can be seen that the removal of tax credits (ITC and PTE) have a slight impact on the DER sizing. When the ITC and PTE are excluded, the capacity of the ESS and PV slightly decreases. This is because both the ITC and PTE are applicable only to the ESS and PV in the configuration. This indicates that the incentives which are designed for a specific technology will increase the penetration of that technology in microgrid systems. The SGIP has a decisive impact on the optimal DER mix. When the SGIP is removed, the FC becomes infeasible, the ESS capacity decreases, and the financial indices are notably worse than that of the optimal PV + ESS configuration. This is because the SGIP provides a significant amount of IBIs, which helps to reduce the installation costs of FC and ESS. This evidence provides insights into the economic importance of renewable energy incentives for microgrid project investment, and they should therefore be included in the design and planning process for microgrids.

Uncertain Financial Parameters
The project financial parameters and the capital costs of DER technologies can vary considerably because they depend on the status of local economy, manufacturers, and technical characteristics. The impact of this uncertainty on the NPV and SPB are examined in this section. Capital costs were altered by between −50% and 50% to reflect existing prices and the downward trend in prices of DERs due to technological innovation. The financial parameters were varied from −25% to 25% because the state of local economy may go up or down. The results of the sensitivity analysis for parameter uncertainty are presented in Figures 12 and 13.
NPV, LCC, SPB, and BCR, which offer various criteria that can be designed for the problem of optimal design. For instance, if an investor requires a minimum period to recover the initial investment capital, SPB will be used as the primary benchmark to select the optimal microgrid configuration. In another case when a microgrid owner asks for a configuration with minimum operating costs, LCC will be selected as the design objective. Therefore, the significance of the proposed methodology is that it can be generalized to different microgrid design problems.  The sensitivity analysis arising from the presented microgrid design process provides a means to examine the importance and influence of incentives on the optimal design. Moreover, this analysis is also helpful for renewable energy policy makers. Indeed, it helps them determine what extent an incentive program or a combination of various types of incentive is effective for promoting more investment in microgrid projects. Based on the results, a policy maker can decide incentive rates, applicable durations, and other incentive strategies which are appropriate for short-term and longterm promotion of renewable energy systems. To increase the penetration of a particular technology, the government or the utility should launch an incentive program specifically designed for that technology with a certain level of incentive rate. Similarly, to promote a specific microgrid configuration, one or more incentive programs should be available to the DERs in that configuration, and a microgrid is eligible for the incentives if, and only if, the specified DERs are integrated together. Furthermore, a tiered structure for incentive rates, which is based on construction time of a microgrid and/or rated capacity of technologies, can be used to encourage the microgrid implementation in certain periods, as well as to support a certain scale of microgrids, such as buildings, campuses or residential areas.

Conclusions
This paper presented a design method for commercial microgrids using MDSTool, which was applied to determine the optimal design for a university campus microgrid. The optimal microgrid configuration was obtained based on financial feasibility analysis considering various renewable energy incentives. With the objective of maximizing profitability by microgrid deployment, incentives and tax benefits were found to have a strong effect on the optimal microgrid design, including the optimal mix and sizing of DERs. In particular, investment-based incentives were found to have a decisive impact because they help to decrease the high installation costs of DERs, which has been the main challenge of integrating hybrid renewable energy systems. Therefore, it is concluded Of the financial parameters, only the inflation rate has a positive influence on the NPV, and the discount rate is the most sensitive. When the discount rate is decreased by 25%, the NPV goes up to $1,887,900.00, while it is $1,376,000.00 if the discount rate is increased by 25%. The loan rate and federal tax rate are the next most sensitive parameters. Compared to the federal tax rate, the state tax rate has a slightly negative impact on the NPV because its nominal value is just 8.84%. When changing from −25% to 25%, the federal tax rate makes the SPB vary from 4.95 to 5.49 y, while the discount and loan rates have no impact on the SPB. The SPB changes marginally around 5.20 y corresponding to variations in the inflation rate and state tax rate.
In terms of capital costs, it is observed that this cost of FC dominantly affects both the NPV and SPB. When the capital cost of FC is increased by 50%, the NPV sharply decreases to $811,340.00, with the SPB rising up to 7.97 y. When it is reduced by 50%, the project is noticeably profitable, with an NPV of $2,376,600.00 and an SPB of only 2.50 y. This can be ascribed to the operation of FC, which results in most of the electricity savings by the microgrid deployment. The next most sensitive capital cost is that of DSM. The costs of ESS and PV have a slight impact, which is owing to their small capacity in the optimal configuration. These sensitivity results illustrate that investment in microgrids will be more financially attractive in the future if the installation costs of DERs continue to fall and more renewable energy incentives are provided by governments or utilities.
The analysis performed above for the university campus microgrid validates the advantages of the proposed microgrid design method using MDSTool. Compared to previous studies, the design approach can evaluate the financial feasibility of a whole microgrid project, rather than of a single DER technology. The techno-economic analysis incorporates the economic benefits from various incentives and tax credits for renewable energy technologies. The problem of optimal mix and sizing of DERs can be solved by defining the design objective function. The results from MDSTool help to intensively evaluate the technical and economic efficiency of a hybrid microgrid system, including the operational efficiency of each energy generation or storage unit, and the performance of an economic dispatch strategy. In addition, the results provide multiple economic measures such as NPV, LCC, SPB, and BCR, which offer various criteria that can be designed for the problem of optimal design. For instance, if an investor requires a minimum period to recover the initial investment capital, SPB will be used as the primary benchmark to select the optimal microgrid configuration. In another case when a microgrid owner asks for a configuration with minimum operating costs, LCC will be selected as the design objective. Therefore, the significance of the proposed methodology is that it can be generalized to different microgrid design problems.
The sensitivity analysis arising from the presented microgrid design process provides a means to examine the importance and influence of incentives on the optimal design. Moreover, this analysis is also helpful for renewable energy policy makers. Indeed, it helps them determine what extent an incentive program or a combination of various types of incentive is effective for promoting more investment in microgrid projects. Based on the results, a policy maker can decide incentive rates, applicable durations, and other incentive strategies which are appropriate for short-term and long-term promotion of renewable energy systems. To increase the penetration of a particular technology, the government or the utility should launch an incentive program specifically designed for that technology with a certain level of incentive rate. Similarly, to promote a specific microgrid configuration, one or more incentive programs should be available to the DERs in that configuration, and a microgrid is eligible for the incentives if, and only if, the specified DERs are integrated together. Furthermore, a tiered structure for incentive rates, which is based on construction time of a microgrid and/or rated capacity of technologies, can be used to encourage the microgrid implementation in certain periods, as well as to support a certain scale of microgrids, such as buildings, campuses or residential areas.

Conclusions
This paper presented a design method for commercial microgrids using MDSTool, which was applied to determine the optimal design for a university campus microgrid. The optimal microgrid configuration was obtained based on financial feasibility analysis considering various renewable energy incentives. With the objective of maximizing profitability by microgrid deployment, incentives and tax benefits were found to have a strong effect on the optimal microgrid design, including the optimal mix and sizing of DERs. In particular, investment-based incentives were found to have a decisive impact because they help to decrease the high installation costs of DERs, which has been the main challenge of integrating hybrid renewable energy systems. Therefore, it is concluded that accessing all associated incentives is crucial for effective microgrid design and planning.
The results of financial feasibility analysis are beneficial to different sectors such as utility, policy makers in renewable energy development, and industry. The California market was investigated as a case study, but the proposed method can be applied to any particular area with different market conditions and incentive policies. The results indicated that investment in microgrid in a market with diversity of incentives for DER technologies will be financially attractive to commercial investors. In some countries where there are few incentive programs for microgrids, or the penetration of RESs is inadequate, this study provides a reference for the government and utility to establish incentive policies for the adoption of more renewable energy systems. Recently, many countries have set their targets for renewable energy generation to significantly reduce emissions from the power sectors. This study highlights that financial incentives for DERs play an important role in achieving renewable and sustainable energy goals.