For this study, isolated grids operated by the National Operator NPC-SPUG, where data are available, were considered. Due to the archipelagic structure of the Philippines, it is costly and in many cases even not technically feasible to connect all demand clusters to the national grid. These isolated grids are mainly relying on diesel power generation, which implies high generation and operational costs due to varying fuel costs and high transport costs. As a paradigm shift, it is attractive to implement renewable energies in these island grids first and to gain experiences and learnings, which could be transferred to the larger Philippine grids later.

A government agency, NPC-SPUG is the largest operator of off-grid systems in the Philippines. In particular 215 NPC-SPUG off-grid systems, located in Luzon (139), Visayas (55), and Mindanao (21), are studied for their potential of implementing renewable energies. These off-grid systems vary in sizes. The peak demand of the three largest grids exceeds 10 MWp, followed by 20 grids in a range between 1–10 MW_p, and the remaining grids have peak capacities of several 100 kW_p down to 4 kW_p.

A Python-based energy systems simulation tool was utilized to optimize the least cost hybrid combination of PV, battery, and diesel capacity for each of the 215 micro-grid/mini-grids operated by NPC-SPUG. The tool has been successfully used in several studies and is validated using Homer Energy [8], [22]. It simulates the energy flow in hourly time steps for one reference year, looking at the energy generated from solar PV and DPP, energy stored to/discharged from the battery, and energy demand of the grid, as derived from the solar insolation data and techno-economic input parameters (see Figure 1).

Figure 1.

Scheme of the island energy supply model indicating the energy flows within the system

The energy flows were calculated using the dispatch strategy displayed in Figure 2, with the grid load requirement to be supplied from solar power, battery charge, and/or diesel power. At each hour for the entire reference year, the energy systems model applies the dispatch strategy of maximizing the amount of energy coming from the solar power plant. First, the model checks if the load in the island can be supplied entirely by solar power. If it is, it then verifies if the stability requirement (rotating mass), which is set at 40% of the island’s energy demand, is followed. With the system always operating at least 40% of the power coming from the battery or diesel generator, the stability is ensured. If charged, the battery should be discharged first before turning on the diesel generator.

Figure 2.

Dispatch strategy of the energy systems model used in the study

On the other hand, if the solar power generated is insufficient to supply the load, the system discharges the battery. Once the battery is fully discharged to its allowable State-of-Charge (SOC), which is 20% in the case of Li-ion battery, the diesel generator is operated. In any case, the stability criterion must be fulfilled by the battery or the diesel generator.

Looking at the available data from NPC-SPUG, the DPPs rated capacities were assumed to be twice that of the peak loads for the island grids. It was assumed that no additional installations of DPPs or replacement of old capacities are to be done during the project lifetime of 20 years. The variable parameters in the study are the size of solar PV and Battery Energy Storage System (BESS). By changing the solar PV and BESS share of the hybrid system, the simulation tool minimizes the LCOE under the constraints imposed by the equations discussed in Short and Packey [23]. The simulations are automated to run independently and simultaneously for each small island grid location.

The location of the individual DPPs were gathered from the NPC-SPUG website^† For a single DPP supplying electricity to a mini-grid, the actual GIS location was taken. Otherwise, the centroid of the multiple DPPs connected to a single micro-grid was taken as the location.

A normalized load profile was used for the simulations. This load profile is based on an average daily load profile, which was derived from the actual data of 22 island grids [24]. The applied daily load profile is characterized by a peak demand in the evening hours and slightly higher demand during the afternoon hours (see Figure 3). The minimal load never falls below 50% of the peak load. These characteristics (low demand during the day, peak demand in the evening) are typical for rural consumers in the Philippines. The monthly variations in energy demands according to [24] were incorporated. Annual variations are quite low with peak values in June reflecting additional demands for cooling appliances, as shown in Figure 3. A generalization of load profiles was necessary for this study as not for each of the NPC-SPUG areas real load profiles were available. The daily load profile was normalized according to the peak demand during the day while the annual load profile was normalized according to the average load during the entire year. It is certain that a number of the considered grids are not operated 24/7 hours. Nevertheless, for this study, it is assumed that all future hybrid systems provide uninterrupted electricity supply to reflect the development goals of the government of the Philippines. For scaling the normalized load profile to each of the considered island grids, peak demand was defined as 50% of the installed power generation capacity in place.

Figure 3.

Normalized daily load profile applied for the simulation (top figure) and normalized annual load profile applied for the simulation (bottom figure)

Data on the Global Horizontal Irradiation (GHI) were gathered from global databases in order to quantify the solar PV potential for all island grids considered (see Figure 4). The NASA database covering the period 1984–2005 was used [25]. To get the GHI time-series at a spatial resolution of 0.5° by 0.5°, the method developed by the German Aerospace Center (DLR), where they applied a clear sky index approach and accounted for hourly changes in the irradiance data, as adopted. In the modelling, the resource pixel of the DPP location or the centroid for multiple DPPs in a single grid was chosen. To translate the irradiation data to actual electricity production, a model using crystalline silicon PV modules was utilized [26]. The effect of daily average temperature and irradiation on the module efficiency was considered following previous works [27], [28]. Using our previous approach, a degradation rate of 0.3% per year and a power reduction factor of 3% to account for the adverse effect of clouding and pollution were applied. When the various models are combined, the system generates the island-based PV yield in hourly time steps.

Figure 4.

Map of the Philippines showing solar PV resource in terms of the GHI

To complete the simulation parameters, the technical and economic data of the Philippine small island grids were loaded into our energy systems model, as shown in Table 1. From previous experiences, diesel fuel cost influences the most the power generation cost of the hybrid system. For the diesel cost, an average value of USD 0.90 per liter across all island grids considered was used, which was taken from the average landing cost of diesel in NPC-SPUG islands [24]. A diesel price inflation of 3.8% per year was used. While the national average price is lower, the landing cost of diesel typically rises due to the transport cost in bringing the diesel by boat to these islands. As discussed later in the sensitivity analysis section, this study also looked into the effect of the diesel prices on the LCOE in each island grid.

Since the DPPs exist already in the NPC-SPUG small islands and are verified to be in operation, it was assumed that there would be no initial cost for the diesel technology. The operational costs, however, which are based on expenditures such as maintenance and lubricant oils, account to around 0.05 USD/kWh [29]. The lifetime of the DPPs was set to be at 20 years. To calculate the efficiency of the DPP as the actual loading changes, the efficiency values described by [30] was used, which were between 30% and 40%. To maintain the stability of the off-grid system, a rotating mass of 40% was set to ensure enough diesel or battery capacities to maintain frequency and voltage control [31].

The solar PV power plant is economically defined by the initial cost or Capital Expenditure (CAPEX), Operational and Maintenance Expenditures (OPEX), and lifetime. Typically, turn-key PV plants in the Philippines costs around 1,200 USD/kW_p and the OPEX is at 25 USD/kWh/yr for a lifetime of 25 years. On the other hand, the Li-ion battery-based BESS was assumed to have an initial cost of 260 USD/kWh for capacity and 540 USD/kW for power, combining the costs at a fixed C-rate of 0.5 to 800 USD for both as a modular unit. OPEX is set at 7.25 USD/kWh per year and a lifetime of 15 years. For the cycle efficiency, a conservative estimate of 90% charging and discharging efficiency, leading to a roundtrip efficiency of 81%, was used.

To initiate the hybridization process, a fixed development cost of 20,000 USD per small island grid was assumed for a project lifetime of 20 years. The Weighted Average Cost of Capital (WACC) was taken to be 9.94% for a typical energy and utility project in the Philippines. Table 1 shows the summary of the technical and economic input parameters loaded into the energy systems model.

For studying the effect of battery costs on the LCOE of hybrid energy systems, the techno-economic sensitivity modelling and optimization was performed across all the surveyed island grids on four scenarios with different battery costs. To generate the four scenarios, the battery investment costs was decreased (−50%, −25%) and was increased (+25%, +50%) relative to the original battery costs in the calculations discussed above. All other parameters were fixed as in the main scenario (see Table 1). The applied values for the sensitivity analysis are listed in Table 2.

Hybridization of the islands avoids the importation of approximately 92 million liters of diesel per year. At a diesel price of 0.9 USD/L, this corresponds to a diesel importation cost of 82 million USD (4.3 billion PhP) annually. For the diesel only case, it was estimated that the annual total importation cost to operate the existing DPPs is around 166 million USD (8.6 billion PhP) annually. Recently, there was a proposal from the Philippine Department of Finance (DOF) to increase the excise tax on diesel by PhP 6 per liter [33]. Surely, this move would impact the future UCME resulting from the increase in the TCGR in these island grids.

Aside from the economic benefits of energy transition, hybridization enables the Greenhouse Gas (GHG) emission savings of around 246 thousand metric tons of CO₂ annually. Luzon (45.6 million L/yr) has the biggest diesel savings potential, followed by Mindanao (31.8 million L/yr) and Visayas (14.5 million L/yr). For the above things to happen, a capital infusion of around 529 million USD is needed (see Table 4).

Recent works argue that not all RE megawatts are created equal [34], [35]. They propose that RE power generation could be strategically used to promote social outcomes where it is really needed (i.e. off-grid rural areas). While the Philippines in recent years has a good track record in increasing the RE share with new solar, wind, and hydro power plants constructed and planned, smaller RE based project would never come into fruition if only megawatts were the only consideration. There is a bias for large, grid-connected facilities. The availability of electricity in the off-grid areas should provide an impetus for an increase and economic activity and production of high value good in the countryside.

For studying the effect of diesel costs on the LCOE of hybrid systems, a techno economic sensitivity modelling and optimization for three scenarios of RE penetration across all the surveyed island grids was performed. A high RE penetration scenario corresponding to an average 60% RE share, a medium RE penetration scenario corresponding to an average 45% RE share, and a low RE penetration scenario corresponding to an average 30% RE share. As shown in Figure 7, at low diesel prices (below 0.4 USD/L, it is expected that the diesel generation only system has the lowest LCOE, followed by the low, medium, and high RE shares. On the other hand, at diesel prices in between 0.5 to 1.0 USD/L, the 30% RE share configuration is the most financially attractive. As diesel prices rise further, more RE penetration decreases the LCOE. For instance, if the diesel prices reach 2.0 USD/L, a 60% RE penetration configuration becomes the cheapest option. While the odds are low in the near future that diesel prices reach that price, a high RE share has the most price stabilizing effect among all options considered. While having a higher LCOE due to the increased capital expenditure on the solar power plant and the Li-ion BESS, the 60% RE scenario is the least dependent on the diesel price, not to mention the avoidance of importing millions of liters of diesel each year and the associated GHG emissions.

Figure 7.

Sensitivity analyses for diesel fuel price at different hybrid system configurations according to the optimized RE shares

The sensitivity analysis reveals the impact of battery costs on the overall LCOE reduction potential and the cost optimized hybrid system design (see Table 5). As expected, decreased battery costs (−50% and −25%) would result in lower LCOE and higher RE shares. Decreased battery costs of −50% lead only to a marginal increase in the RE share (+0.16%) compared to the −25% battery cost scenario. The applied stability criteria (40% rotating mass) and diesel price prevent a higher RE share and a higher battery potential. Further, decreasing battery and PV investment costs in combination with higher initial diesel costs or a higher annual diesel price growth could change this pattern.

A remarkable “tipping point” can be observed when comparing the base scenario to the +25% battery cost scenario. With increasing battery costs, the battery potential shrinks significantly (−70%) and the RE share decreases from 50.4% to 37.1%. In contrast to the battery potential, the PV potential is decreased by 21%. Thus, with the increased battery costs of +25%, it is less economically feasible to implement battery capacities in the hybrid system to increase the overall RE share, however, it is still economically feasible to implement PV capacities for substituting diesel power generation. This “tipping point” reflects the economic breakpoint for hybrid “fuel saver” systems (diesel and PV) to hybrid battery systems (diesel, PV and battery).

With regard to the LCOE of the hybrid systems the sensitivity analysis reveals that changing battery costs would only slightly affect the overall power generation costs. Lower battery costs (−50%) result in an additional average hybrid LCOE reduction potential of 0.03 USD/kWh whereas higher battery costs (+50%) increase the average hybrid LCOE by 0.01 USD/kWh.

Overall, the sensitivity analysis highlights that even when applying high battery costs (+50%) a significant LCOE reduction potential (17.6%) compared to the diesel only system exists. Nevertheless, with increasing battery costs the cost optimized RE share and battery potential decreases significantly. Recent papers argue that the cost of battery energy storage systems for microgrids will continue to decrease due to economy of scale and continuous deployment of grid-scale battery technologies and ramping up in R&D investments [13], [36].

Continuing on the reliance on diesel in the future will surely lead to a rise in missionary electrification subsidies. The steeper slope of the “diesel only” case emphasizes the heavy dependence of the LCOE to the diesel price in the worldwide market. For instance, according to the first official draft of the Missionary Electrification Development Plan (MEDP) of the DOE for the period 2017–2021 [5] the total UCME subsidy will increase from 138.5 million USD (7.2 billion PhP) in 2015 up to 273.1 million USD (14.2 billion PhP) in 2021. In 2016, the distribution utilities were collecting an average UCME of PhP 0.19/kWh to all electricity consumers in the country.

For the planning period 2017–2021 considered above, missionary electrification in the Philippies requires a total subsidy of 1.463 billion USD (76.06 billion PhP). Additionally, DOE expects the addition of around 111.7 MW of installed capacity in the NPC–SPUG areas, which could stretch further the UCME requirements. While there are on-going privatization program in the NPC–SPUG areas in recent years, the reduction of costs and improved operational efficiency are not yet sufficient to significantly reduce or even remove the UCME subsidies. There are already fifteen NPPs with a total rated capacity of around 206 MW powering nine off-grid areas with 24-hr electricity service. Most of the generation plants, however, are still based on conventional fossil fuels like coal, natural gas, and diesel. Yet, even with the power generation based on GHG-emitting plants, there is still a need to infuse the gap between the true cost and the subsidized generation rates by the missionary electrification subsidy.

Although the electricity consumers in the NPC–SPUG areas, who are mainly in the marginalized sectors, pay a subsidized rate, this dependence on fossil fuels makes electricity more expensive to everyone because UCME is eventually collected to all grid-connected consumers. Taking into account the identified LCOE reduction potential of 19.8% on average, our study underlines the feasibility of reducing the UCME charge through investing in renewable energies. As emphasized above, dependence on diesel has its associated externalities costs such as GHG emissions and environmental pollution. Aside from improving the efficiency of the diesel-fired power plants, increasing the penetration of renewable energy in missionary areas to rationalize the UCME subsidy should be a key step going forward.