Cost Analysis of Optimized Islanded Energy Systems in a Dispersed Air Base Conﬂict

: The United States Air Force has implemented a dispersed air base strategy to enhance mission e ﬀ ectiveness for near-peer conﬂicts. Asset dispersal places many smaller bases across a wide geographic area, which increases resupply requirements and logistical complexity. Hybrid energy systems reduce resupply requirements through sustainable, o ﬀ -grid energy production. This paper presents a novel hybrid energy renewable delivery system (HERDS) model capable of (1) selecting the optimal hybrid energy system design that meets demand at the lowest net present cost and (2) optimizing the delivery of the selected system using existing Air Force cargo aircraft. The novelty of the model’s capabilities is displayed using Clark Air Base, Philippines as a case study. The HERDS model selected an optimal conﬁguration consisting of a 676-kW photovoltaic array, an 1846-kWh battery system, and a 200-kW generator. This hybrid energy system predicts a 54% reduction in cost and an 88% reduction in fuel usage, as compared to the baseline Air Force system. The HERDS model is expected to support planners in their ongoing e ﬀ orts to construct cost-e ﬀ ective sites that minimize the transport and logistic requirements associated with remote installations. Additionally, the results of this paper may be appropriate for broader civilian applications.


Introduction
After nearly two decades of counterterrorism and counterinsurgency, the Unites States (US) Department of Defense (DoD) 2018 National Defense Strategy (NDS) has shifted its primary focus to near-peer conflict [1]. In an armed conflict with near-peer adversaries, air superiority will be contested; therefore, forces may need to be on constant alert within the contested airspace. To mitigate risk in a near-peer conflict, the RAND Corporation proposed that the Air Force employ a dispersed air base concept that involves conducting operations from many smaller air bases [2]. This strategy will allow the Air Force to set up and operate in one location and rapidly move operations before the base is targeted for attack. However, conducting operations from a large number of bases inherently increases the fuel, munitions, food, and water resupply logistical complexity and its exposure to disruptions by adversaries [3].
The RAND report focused on the possibility of a conflict in the area surrounding the South China Sea, and it identified that a mixture of different types of bases is required to achieve military objectives in a contested battlespace-"stay-and-fight" may have an indefinite duration, "drop-in" may last days to weeks, and "fighter forward arming and refueling point (FARP)" may be established for less than a day. Stay-and-fight bases have a larger footprint, operate on the edge of the contested battlespace, and have more permanent infrastructure, including grid-tied power and large-scale fuel storage. One the other end of the spectrum, fighter FARP bases are conceptualized as austere airfields, where everything needed to conduct operations would be flown in on cargo aircraft and only used for short

Materials and Methods
The hybrid energy renewable delivery system (HERDS) model was constructed to optimize a HES system to increase the energy resiliency of a base and to reduce the amount of fuel it consumes by generating power. The HERDS model is formulated in the three distinct stages shown in Figure 1: the data inputs needed to run the model, the calculations performed from the input parameters, and the resulting output values and configurations. In the input stage, the site location and number of personnel stationed at the base are needed, in addition to the physical, performance, and cost attributes of the different components in the HES. During the calculations stage, an electric load profile is generated and used to compute the optimal HES combinations that can then be airlifted with a select combination of aircraft. The final outputs of the model are the configurations of the HES, the airframes needed for system transportation, and a total cost in present dollar terms that can be used to compare alternative courses of action.
Utilizing the HOMER software simplifies this process. Once a location is selected, the software downloads the appropriate solar and wind resource data from the National Renewable Energy Lab (NREL) database, as well as the National Aeronautics and Space Administration (NASA) Surface Meteorology and Solar Energy database [29]. This data is then used to predict the amount of energy produced and stored by the microgrid at a specific location. The different HES components, such as solar panels, wind turbines, energy storage, and inverters, can be added and edited individually to replicate a real-world system. Figure 2 displays a generic HES architecture that can be used within HOMER. The software also allows the user to specify the economical parameters of the model as a whole and for each HES component. This architecture is evaluated within HERDS and, also, two variations of the architecture-either without the solar array or without the wind turbine. An overview of the input parameters, calculations performed, and outputs of the hybrid energy renewable delivery system (HERDS) model [31]. HES: hybrid energy system and NPC: net present cost.
Utilizing the HOMER software simplifies this process. Once a location is selected, the software downloads the appropriate solar and wind resource data from the National Renewable Energy Lab (NREL) database, as well as the National Aeronautics and Space Administration (NASA) Surface Meteorology and Solar Energy database [29]. This data is then used to predict the amount of energy produced and stored by the microgrid at a specific location. The different HES components, such as solar panels, wind turbines, energy storage, and inverters, can be added and edited individually to replicate a real-world system. Figure 2 displays a generic HES architecture that can be used within HOMER. The software also allows the user to specify the economical parameters of the model as a whole and for each HES component. This architecture is evaluated within HERDS and, also, two variations of the architecture-either without the solar array or without the wind turbine. An overview of the input parameters, calculations performed, and outputs of the hybrid energy renewable delivery system (HERDS) model [31]. HES: hybrid energy system and NPC: net present cost. The electrical load requirements must be known to model a site and generate an optimally sized HES. The primary facility type used in drop-in bases are fabric shelters. The Alaska Small Shelter System, shown in Figure 3, is representative of this type of facility. It consists of hollow aluminum segments held together by rack and pin placed directly on the ground; both the floor and exterior shell consist of a fabric liner. The exterior shell material, measured as 1.6-mm-thick, is the primary barrier between the exterior environment and the air conditioned space within the shelter [32]. System, shown in Figure 3, is representative of this type of facility. It consists of hollow aluminum segments held together by rack and pin placed directly on the ground; both the floor and exterior shell consist of a fabric liner. The exterior shell material, measured as 1.6-mm-thick, is the primary barrier between the exterior environment and the air conditioned space within the shelter [32]. The electrical load requirements must be known to model a site and generate an optimally sized HES. The primary facility type used in drop-in bases are fabric shelters. The Alaska Small Shelter System, shown in Figure 3, is representative of this type of facility. It consists of hollow aluminum segments held together by rack and pin placed directly on the ground; both the floor and exterior shell consist of a fabric liner. The exterior shell material, measured as 1.6-mm-thick, is the primary barrier between the exterior environment and the air conditioned space within the shelter [32]. One load that all of these shelters share is the power required to maintain the shelter at a stable temperature-typically, 21 °C. An environmental control unit (ECU) is attached to every fabric shelter in the base. Cumulatively, shelter heating and cooling demands are responsible for the majority of the encampment's energy consumption [32]. Data from metering demonstrations at test sites have revealed a relationship between the external ambient temperature and the load required to heat or cool the shelter [4]. The fabric shelter differs significantly from permanent construction envelopes, where internal heat gains are much more notable. The tent's envelope is drastically thinner, resulting in conduction and convection being the driving heat transfer modes. This relationship can then be applied to heat transfer equations to determine the power needed for the ECU to maintain the desired temperature within a tent, as shown in Equation (1) [34,35]. One load that all of these shelters share is the power required to maintain the shelter at a stable temperature-typically, 21 • C. An environmental control unit (ECU) is attached to every fabric shelter in the base. Cumulatively, shelter heating and cooling demands are responsible for the majority of the encampment's energy consumption [32]. Data from metering demonstrations at test sites have revealed a relationship between the external ambient temperature and the load required to heat or cool the shelter [4]. The fabric shelter differs significantly from permanent construction envelopes, where internal heat gains are much more notable. The tent's envelope is drastically thinner, resulting in conduction and convection being the driving heat transfer modes. This relationship can then be applied to heat transfer equations to determine the power needed for the ECU to maintain the desired temperature within a tent, as shown in Equation (1) [34,35].
Equation (1). ECU power draw equation. η represents the energy efficiency ratio of the ECU (modeled as 1.69 W heat/cool W elec ) [36]; A t m 2 is the exposed surface area of the tent; T o ( • C) is the outside air temperature; T i ( • C) is the inside air temperature; and U W m 2 · • C is the overall coefficient of heat transmission, including the air films, tent material, and insulation. Added as a base load requirement is 2 kW to run the ventilation fan. The 3 is a constant to account for additional heat transfer through radiation and air infiltration [4]. One thousand is used to convert the units from Watts to kW.
Equation (1) is used to account for the thermal resistive effects of the different layers between the exterior and the interior environment of the shelter. The U-value (Equation (2)) sums the resistive element between the ambient temperature (T o ) and the interior temperature (T i ) to account for the changes in the heat flow across the different materials, accounting for their thickness and thermally conductive properties. Figure 4 displays the relationship between the interior and exterior temperatures concerning heat flow across the various material layers of the tent.
Equation (2). Cumulative heat transmission coefficient, where x i is the thickness of the physical layer, and k i is the thermal conductance of the material.  (2) Equation (2). Cumulative heat transmission coefficient, where xi is the thickness of the physical layer, and ki is the thermal conductance of the material. In addition to the electrical load from the ECU, the tents draw power from the system for other requirements. Some of these loads are broken down by the tent function, either a billeting tent or a mission-oriented shelter. Billeting tents are where the occupants rest, and mission-oriented shelters house the array of sensors and communication equipment used by the base and are manned 24 h per day. Metering reports from Afghanistan show that other loads occur at regular intervals during the day, such as when the lights come on in a billeting tent or when a new shift starts in the missionoriented tents, and make up approximately 0.2% of the overall camp load [37]. The combination of all the loading factors is listed in Table 1. In addition to the electrical load from the ECU, the tents draw power from the system for other requirements. Some of these loads are broken down by the tent function, either a billeting tent or a mission-oriented shelter. Billeting tents are where the occupants rest, and mission-oriented shelters house the array of sensors and communication equipment used by the base and are manned 24 h per day. Metering reports from Afghanistan show that other loads occur at regular intervals during the day, such as when the lights come on in a billeting tent or when a new shift starts in the mission-oriented tents, and make up approximately 0.2% of the overall camp load [37]. The combination of all the loading factors is listed in Table 1. Before HES-sizing can occur, HOMER requires the user to either select from a set of generic components or input the system performance data, as well as the purchase and operational costs associated with the equipment. After the microgrid architecture has been established and the annual loading requirement is defined, the software then runs simulations for every combination of components (diesel generator only, diesel generator and a battery system, PV and a battery system, etc.) and computes the optimized system for each combination using the net present cost (NPC) with an inflation rate of 2.00%. The NPC considers the equipment purchase cost, the costs for component replacements, the system maintenance cost, and generator fuel consumption cost. Costs other than the initial cost are converted to a present value, which allows for projects of different sizes, lifecycles, and geographic locations to be compared.
HOMER is not able to incorporate transportation costs into its optimization calculations. To account for these costs, the optimal HES is converted into pallet position equivalents. Using the aircraft selection model, the cheapest transportation option is selected, and a total airlift cost is calculated based on the flight time between the staged material's location and the drop-in base site [30].
Calculating pallet position equivalents is dependent on the type of resource required. Using the Air Force standard 463-L pallet constraints, the resource can be divided up either by weight or volume. The maximum weight for one pallet position is 4535 kg (10,000 lbs.), and the maximum volume for one pallet is limited to 2.74-m-high by 2.24-m-long by 2.44-m-tall [38]. The best combination of aircraft needed to airlift the HES to the base is determined using the resulting pallet equivalents. The most common types of aircraft used for airlift operations are the C-130J-30, C-17A, and the C-5A, and their basic specifications are listed in Table 2 [30]. Using the aircraft listed in Table 2, the ASM then determines the number and type of aircraft required to airlift the system. To do this, the model assumes that all aircraft are equally available for transport. First, it calculates all possible combinations of aircraft that can transport the palletized HES by weight and compares them against each aircraft's maximum payload specifications. The model then compares the cargo requirement against the amount of pallet positions that are available within each of the proposed aircraft to determine each applicable airlift combination. Using each possible airlift combination, the model calculates the cumulative cost per flight hour. This cost is primarily derived from the fuel usage rate of each aircraft, as well as the total ownership costs, which include the cost of purchase, maintenance, and repairs for the aircraft [30]. This cost per hour is used to determine the least expensive option for transporting the HES to its destination. The total flight cost, as a function of the total flight time and cost per hour, is then added (without discounting) to the net present cost of the HES.

Results
To demonstrate the unique capabilities of the HERDS model, a case study was analyzed utilizing a drop-in base with 300 assigned personnel at Clark Air Base, Philippines. This size base was chosen because the personnel there will be using expeditionary equipment and will likely utilize fabric shelters [2]. Given the camp population and facility requirements, access to prepositioned fuel sources is unlikely, and leadership will have to prioritize the fuel they have available to primarily support flight line operations. These traits make a drop-in base of this size an ideal situation to integrate a HES.
Several assumptions are made to simulate the battlefield circumstances accurately. First, all required material for a HES is prepositioned at three nearby staging countries with stay-and-fight-sized bases: Japan, Guam, or Australia. Next, it is assumed that all required austere location base equipment (tents, generators, electrical equipment, etc.) are already in place and need no further material from additional airlifts. To simulate the existing power grid, a 200-kW generator was added to the model to generate any power that could not be met with PV cells or wind turbines. Since the generator capability is already assumed to be in place, the smallest allowable purchase cost was selected-$0.01/kW. The fuel Energies 2020, 13, 4677 8 of 17 price was set at $2.00 per liter, which includes the cost of the fuel and an estimate of transportation costs to the site.
Additionally, all HES equipment is assumed to be purchased without a loan. In order to represent this within HOMER's NPC calculations, a discount rate of 0.01% was used. Finally, the HES equipment, once transported to Clark Air Base, will continue to be utilized for the duration of operations [2]. A project lifespan of 15 years was used to model the utilization of the components until failure.
Time-series data for the ambient temperature in the area was collected from NASA's global weather data in one-hour intervals and used to generate the initial electric load of the shelters [40,41]. This temperature data is applied to Equation (1) to generate the load of the ECU connected to the tent. For this case study, the modeled shelter uses a 2.54-cm Thinsulate layer of insulation, which has an individual U-value of 1.2 W m 2 · • C and a cumulative heat transmission coefficient of 0.9 W m 2 · • C The calculated ECU load is then added to the other loads listed in Table 1 to generate the estimated daily load profiles for fabric shelters at Clark Air Base, Philippines. Figure 5 displays the resulting loads for one day for both a billeting and a mission tent.
sized bases: Japan, Guam, or Australia. Next, it is assumed that all required austere location base equipment (tents, generators, electrical equipment, etc.) are already in place and need no further material from additional airlifts. To simulate the existing power grid, a 200-kW generator was added to the model to generate any power that could not be met with PV cells or wind turbines. Since the generator capability is already assumed to be in place, the smallest allowable purchase cost was selected-$0.01/kW. The fuel price was set at $2.00 per liter, which includes the cost of the fuel and an estimate of transportation costs to the site.
Additionally, all HES equipment is assumed to be purchased without a loan. In order to represent this within HOMER's NPC calculations, a discount rate of 0.01% was used. Finally, the HES equipment, once transported to Clark Air Base, will continue to be utilized for the duration of operations [2]. A project lifespan of 15 years was used to model the utilization of the components until failure.
Time-series data for the ambient temperature in the area was collected from NASA's global weather data in one-hour intervals and used to generate the initial electric load of the shelters [40,41]. This temperature data is applied to Equation (1) to generate the load of the ECU connected to the tent. For this case study, the modeled shelter uses a 2.54-cm Thinsulate layer of insulation, which has an individual U-value of 1.2 •℃ and a cumulative heat transmission coefficient of 0.9

•℃
The calculated ECU load is then added to the other loads listed in Table 1 to generate the estimated daily load profiles for fabric shelters at Clark Air Base, Philippines. Figure 5 displays the resulting loads for one day for both a billeting and a mission tent.  Single shelter loads were then scaled to reflect the size of a drop-in base. The 300 personnel needed to operate the base were divided among 25 billeting tents and five mission tents [42]. The aggregate daily loads were translated to annual loads using HOMER, as shown in Figure 6. Next, the estimated annual load for the base was used to compute the HES optimal design points. Single shelter loads were then scaled to reflect the size of a drop-in base. The 300 personnel needed to operate the base were divided among 25 billeting tents and five mission tents [42]. The aggregate daily loads were translated to annual loads using HOMER, as shown in Figure 6. Next, the estimated annual load for the base was used to compute the HES optimal design points. For this case study, specific components were chosen due to their market availability, performance specifications, and ability to be transported in pallets aboard aircrafts. The PV panels being modeled in this study are SunPower E20 [43]. They have a rated capacity of 327 W and an efficiency of 20.4%, which is 3-5% more efficient than most commercially available panels [44]. Panel efficiency was a prime consideration for the case study in order to minimize the area required for the array. Table 3 lists additional manufacturer specifications for the solar panel [43,45,46]. For this case study, specific components were chosen due to their market availability, performance specifications, and ability to be transported in pallets aboard aircrafts. The PV panels being modeled Energies 2020, 13, 4677 9 of 17 in this study are SunPower E20 [43]. They have a rated capacity of 327 W and an efficiency of 20.4%, which is 3-5% more efficient than most commercially available panels [44]. Panel efficiency was a prime consideration for the case study in order to minimize the area required for the array. Table 3 lists additional manufacturer specifications for the solar panel [43,45,46]. Wind turbines used in the simulation are representative of the Kingspan KW6 [45], which is rated at 6 kW, has a 15-m-tall tower, and a 5.6-m rotor diameter. Its size makes it ideal for accessing the wind above the structures of the camp, while still being small enough not to interfere with air operations [26]. The turbine's cost data are listed in Table 3.
The energy-storage optimization process was calculated within HOMER using 1-kWh intervals in the standard Li-ion battery. This interval enabled a high level of precision in calculating the needed energy storage. The resulting energy-storage requirement is then converted to an integer number of the 210-kWh Tesla power bank system [46] in order to model the airlift requirements. A summary of the component parameters is presented in Table 3.
Using the components listed in Table 3, HOMER generated 5060 different systems. Each HES listed in Table 4 is the system with the lowest NPC over the 15-year lifespan for each unique architecture. HES 1 and HES 2 resulted in similar-sized arrays, energy storage, and NPC. The main difference in the two systems is that HES 1 only has a combination of solar PV and battery backup, along with a generator, while HES 2 adds a single wind turbine. The addition of a wind turbine caused the NPC to increase slightly for HES 2 and resulted in using 1600-L more fuel per year. Next, the transportation costs of both HES combinations were calculated.
The following criteria were used to divide the HES into pallet equivalents. Photovoltaic panel pallets will be governed by volume due to individual panel weights of 18.6 kg [43]. Racking systems, similar to Sollega solar buckets, were included [47], as well as individual microinverters, and all needed connection equipment. Each photovoltaic panel pallet had a rating of 40 kW.
The Kingspan KW6 s 15-m tower [45] can be divided into different sections to meet the dimensional requirements of a pallet and later bolted together on-site. Due to weight requirements, one entire system was split between two aircraft pallets. One pallet contained the base and tower sections, while the other pallet contained the wind turbine and blades. Altogether, the pallet combination had a rated capacity of 6 kW [45].
Energy-storage systems are generally heavier than most other HES components. Systems such as the Tesla Powerpack can store up to 210 kWh and weigh 1622 kg; thus, only two battery banks can be transported per pallet [46]. Table 5 displays the details and contents of each pallet type. The resulting pallets were then divided up among the different airframes listed in Table 2. The ASM was employed by using the volume (pallet positions) and weight (kg) of all cargo aircraft, then creating all possible combinations of smaller airframes and checking for the lowest cost per flight hour. Table 6 displays all combinations of aircraft capable of transporting the HES and their cost per flight hour. In the table below, HES 1 consists of 22 pallets and 57,352 kg, while HES 2 consists of 24 pallets and 64,610 kg. Each row in Table 6 is a different combination of aircraft sorties that can transport either system. The optimal aircraft combinations for HES 1 and 2 differed primarily due to HES 2 including a 7250-kg wind turbine. The wind turbine caused the needed pallets to exceed the maximum payload of three C-130J-30s by 4900 kg. Without the turbine, HES 1 was able to fit within three C-130J-30s and result in a lower cost per flight hour.
As stated in the RAND report, all the material needed to support these bases is assumed to be prepositioned in the nearby countries of Japan, Guam, and Australia. Figure 7 shows each location in relation to Clark Air Base and the estimated flight duration between each location. As shown in Figure 7, the closest theoretical staging point for Clark Air Base is in Japan. The total estimated flight duration of eight hours and fifty minutes was used to approximate the total transportation cost of airlifting the material to the drop-in base. Table 7 shows the final calculated costs of operating and transporting the HES. As shown in Figure 7, the closest theoretical staging point for Clark Air Base is in Japan. The total estimated flight duration of eight hours and fifty minutes was used to approximate the total transportation cost of airlifting the material to the drop-in base. Table 7 shows the final calculated costs of operating and transporting the HES. Including transportation costs, HES 1 and 2 could avoid $4.7 or $4.6 million of system lifecycle costs over a 15-year lifespan, respectively. When compared to the generator-only solution, HES 1 saves 54.4% in cost and reduces the fuel consumption by 88%. For reference, the generator-only solution (HES 5, Table 4) possessed an NPC of $9.8 million and used 269,477 L of fuel per year.
To account for future climate uncertainty, a sensitivity analysis was conducted to show HES designs and costs using climate-informed loading extremes. This was accomplished by applying the maximum and minimum hourly electrical loads imposed on the system and projecting those loads across the entire day. The resulting systems would identify the possible sizes and cost variations of future systems. The annual load profiles for each case are shown in Figure 8. The maximum electrical load profile had an average of 3148 kWh/day, which is 21% higher than the actual annual load profile from Figure 6. The minimum profile resulted in 2104 kWh/day averaged across the year, which is 19% lower than actual data. The resulting systems size and cost are displayed in Table 8.
Energies 2020, 13, x FOR PEER REVIEW 13 of 17 future systems. The annual load profiles for each case are shown in Figure 8. The maximum electrical load profile had an average of 3148 kWh/day, which is 21% higher than the actual annual load profile from Figure 6. The minimum profile resulted in 2104 kWh/day averaged across the year, which is 19% lower than actual data. The resulting systems size and cost are displayed in Table 8.  The sensitivity analysis demonstrates that the system size and cost are closely related to the average daily load. Both the size of the PV array and the NPC were within three percentage points  The sensitivity analysis demonstrates that the system size and cost are closely related to the average daily load. Both the size of the PV array and the NPC were within three percentage points when compared to the change in average kWh produced per day. At the extremes of the sensitivity range, only small total annual load deviations were observed in comparison to the base case. From a system design perspective, the small deviations generated battery size changes that corresponded with the load increases or decreases. These changes are a result of the system meeting the peak level power throughout the evening hours when supply cannot be supplemented with solar energy. The sensitivity analysis illustrates that a HES can be scaled proportionally to account for anticipated changes in the climate over the duration of a project.
An additional sensitivity analysis was performed for fuel price and base duration to explore the model variability. The base duration is referred to as the project lifecycle in this section. Fuel price was hypothesized to be important to the HES optimization model, and the fuel price was varied from $1-10/L to determine if the optimal solution favored an increased reliance on renewable energy components. As shown in Figure 9, the optimal HES system size increased with the fuel costs; additional fuel savings made it cost-effective to have a larger HES. Additionally, the project lifecycle was varied from one to five years to test the solution sensitivity. Figure 9 also shows the size of the photovoltaic (PV) array and the energy storage (ES) component for the optimal system at each interval in the sensitivity analysis.
The results of the analysis showed an increasingly linear relationship over time with an increasing fuel price to an increase in size and cost of the HES. This near-linear relationship between the NPC and lifecycle time can be demonstrated by the top-left graph in Figure 9. Each system NPC for every fuel cost iteration was divided by the project duration, resulting in the same increasing rate of cost for every project lifespan. In all cases, the optimal architecture was a combination of solar panels, a backup battery system, and a generator. This result only reinforces the conclusion that the system configuration HES 1 from Table 4 is the best HES combination for Clark Air Base.
One other aspect that became apparent during the sensitivity analysis is that the components' salvage value was artificially deflating the results. Salvage value refers to the amount of money that could be received for selling a component at the end of the project lifecycle. A large portion of the overall NPC was due to the salvage value and did not accurately reflect the system's lifecycle cost. As an example, a 2-MW solar array costing $2.8 M has an expected lifespan of 25 years. If the overall project has a lifespan of two years, the PV panels will have only lost 2/25ths of their initial value and can be sold to recoup costs; this is reflected by an NPC that is significantly lower than the $2.8 M initial cost. The salvage value term within the NPC calculation does not reflect how the military handles its assets. Once purchased, the military will continue to use the components until they are upgraded or no longer functioning. The final project duration was set to 15 years to mitigate this term. This point in time is when the first component (Li-ion battery pack) within the HES architecture is due to be replaced. By ending the project after 15 years, any salvage value incurred will be offset by the replacement cost of the battery pack.  The results of the analysis showed an increasingly linear relationship over time with an increasing fuel price to an increase in size and cost of the HES. This near-linear relationship between the NPC and lifecycle time can be demonstrated by the top-left graph in Figure 9. Each system NPC for every fuel cost iteration was divided by the project duration, resulting in the same increasing rate of cost for every project lifespan. In all cases, the optimal architecture was a combination of solar panels, a backup battery system, and a generator. This result only reinforces the conclusion that the system configuration HES 1 from Table 4 is the best HES combination for Clark Air Base.
One other aspect that became apparent during the sensitivity analysis is that the components' salvage value was artificially deflating the results. Salvage value refers to the amount of money that could be received for selling a component at the end of the project lifecycle. A large portion of the overall NPC was due to the salvage value and did not accurately reflect the system's lifecycle cost. As an example, a 2-MW solar array costing $2.8 M has an expected lifespan of 25 years. If the overall project has a lifespan of two years, the PV panels will have only lost 2/25ths of their initial value and can be sold to recoup costs; this is reflected by an NPC that is significantly lower than the $2.8 M initial cost. The salvage value term within the NPC calculation does not reflect how the military handles its assets. Once purchased, the military will continue to use the components until they are upgraded or no longer functioning. The final project duration was set to 15 years to mitigate this term. This point in time is when the first component (Li-ion battery pack) within the HES architecture

Conclusions
This paper presented the innovative HERDS model to design a hybrid energy system in support of an expeditionary base in a contested environment. The HERDS model is capable of minimizing the transportation and lifecycle costs of a HES, based on a specific climate of the base. A case study was evaluated to highlight the significance and demonstrate the HERDS model's unique capabilities to (1) predict the power requirements of a camp using fabric shelters, (2) design an optimal HES to meet the required load at a minimal operating cost, and (3) account for airlift requirements and costs and assimilate those values into a single cost to be compared against other projects.
The case study was able to quantify the benefits of implementing a HES designed by the model. The standard case of powering a base with a generator had an NPC of $9.81 million, while the best alternative HES had an NPC of $4.99 million, which was 51% of the baseline cost. These savings directly reflect the 237,441 L of fuel saved a year by the HES. An additional savings of $165,000 was also identified by transporting the HES with three C-130J-30s instead of a single C-5 from its staging area to the base's location.
While useful in the design of off-the-grid systems, the HERDS model does possess limitations that could be the subject of future research. The model was developed with military applications in mind. Consequently, the model was tailored to only account for fabric shelters and the primary heat transfer modes by which they are affected. Internal heat and solar gains were determined to be insignificant based on the research and demonstrations that were examined in the literature review. The ASM is currently limited to United States Air Force aircraft and assumes the same speed of flight between the different airframes. The economic model used also differs from civilian acquisition models, because it assumes that the equipment will be purchased upfront and not have a loan to pay back during the lifecycle of the project.
These distinctive capabilities of the HERDS model accurately and efficiently evaluate all feasible design configurations in order to select the optimal HES that minimizes the transportation and lifecycle costs. This model will enable base planners to construct cost-effective, energy-resilient bases, all while reducing the exposed logistical tail.