1 Introduction

High penetration of solar photovoltaic (PV) energy using grid-tied PV system may cause grid transmission problems. Distributed energy generation with energy storage is a good choice to cope with this problem. However, the cost of energy storage is very high and the solar PV system may not be economic.

Hsu et al. [1, 2] developed a hybrid solar PV system (HyPV) as shown in Fig. 1. HyPV operates in stand-alone PV mode or grid mode automatically using switching technique. No solar power is fed into grid. When solar energy is available, including PV power generation and battery storage, it operates in PV mode and the load is powered completely by solar energy as a stand-alone PV. When PV power generation and battery storage is low, it switches to grid mode through ATS (Automatic Transfer Switch). The load is supplied completely by grid, and the battery is charged again by solar PV. HyPV will switch back to PV mode when the battery is charged certain amount of energy.

Fig. 1
figure 1

Schematic diagram of HyPV [1]

Full size image

Solar home system generating electrical power using HyPV with storage device for self-consumption was further proposed [3]. This kind of PV system is usually in small size (1–3 kWp) and looks like a home appliance. Thus, it can be combined with hot water supply system since there is always a hot water demand in a household. Since the thermal storage is much cheaper than battery, this will reduce the cost of energy storage dramatically.

Since no solar PV power is fed into the grid in HyPV, the energy storage has to be large enough in order to avoid the PV energy generation loss. An electric water heater is used to store PV energy as heat to provide enough storage capacity, called “HyPV solar home system” (Fig. 2). In PV mode, the PV energy will be stored as hot water (thermal storage) by the electric heater through ATS2 when the battery is full. Since thermal storage is much cheaper than battery, the dual energy storage would reduce the total cost of HyPV solar home system.

Fig. 2
figure 2

HyPV solar home system with dual energy storage

Full size image

Bocklisch [4] discussed hybrid energy storage (HESS) for power flow control for peak shaving. Lee [5] theoretically studied solar lighting system with battery and heater which stores PV energy at daytime when the battery is full.

Akbaria et al. [6] reviewed all types of energy storage systems that can be integrated with PV system consisting of electrical and thermal energy storage. The integration of PV energy storage in smart buildings is discussed together with the role of energy storage for PV in the context of future energy storage developments. The future energy storage development was summarized, and even demand-side energy management was discussed. It is shown that thermal storage will satisfy thermal comfort needs of buildings when operating with heat pumps while battery deployment may operate with electric vehicles and/or electric heat pumps at times of local grid congestion.

Faccia et al. [7] analyzed the transition from natural gas to electricity-based heating for residential applications, considering the interplay between grid-tied solar PV electricity generation on site and the thermal energy storage to optimize the heating devices. The analysis found that if heat pumps are used instead of natural gas boilers, energy costs are reduced up to 41%. The simultaneous utilization of heat pumps (as substitutes of boilers) and PV panels is shown to be economic and yields grid parity.

The studies of [3,4,5,6,7,8,9] are all related to grid-tied solar PV system with energy storage, and no experimental data were reported. The present HyPV solar home system is basically a stand-alone PV combined with grid power through a switching device. No PV power is fed into grid to cause possible grid congestion. Hence, the solar system requires a larger energy storage to obtain high PV generation efficiency. This may lead to high cost.

The present study focuses on the technical feasibility of non-grid-tied solar home system with dual energy storage. Four HyPV solar home systems with dual energy storage were built for field test to collect the long-term performance data. Solar energy production cost was also analyzed.

2 Installation of HyPV system

2.1 Design of HyPV systems

Four HyPV systems were built in the present study. The system design specification is shown in Table 1. Three of them (D3, D5, and D11) were installed in household to collect data in real application as a home appliance. The PV power is used to drive 2 or 3 air conditioners and LED lighting in the house. The city water is preheated by solar PV power when the battery is full. The 120L water heater with electric heating element provides 4.2 kWh thermal storage capacity (Ew) for water temperature rise ΔT = 30 °C which is larger than the usable battery capacity (Ebat).

Table 1 System design specification of four HyPVs
Full size table

As shown in Table 1, the ratio of battery storage to maximum PV energy generation (Ebat/Epv) for the four HyPVs is all less than 1.0. This will cause solar PV energy generation loss if only battery storage is used. In the present study, the total usual energy storage capacity (Etot) including battery and hot water storage ranges from 5.35 to 7.08 kWh. With this combination, the ratio of total energy storage to expected solar PV energy generation (Etot/Epv) ranges from 1.36 to 2.11. It is known that the FIT solar PV systems generate maximum solar power which is directly fed into grid at optimal energy conversion efficiency since the MPPT (maximum power point tracking) was always used. This means that the designed total energy storage capacity (Etot) including battery and hot water storage is large enough to store the daily solar PV energy generation, if the daily hot water load is larger than 4.2 kWh.

2.2 Installation of HyPV systems

Figures 3, 4, 5 and 6 show the installation of four HyPVs in Taipei and Chia-Yi (southern Taiwan). D3 was installed in a resident house surrounded by apartment buildings. D5 was installed at the top of an 7-floor apartment building. D8 was installed at the solar energy laboratory, National Taiwan University, on the rooftop of a building. D11 was installed in Chia-Yi, southern part of Taiwan, in a residential house. Only D5 utilizes Li battery in the dual energy storage. D3, D8, and D11 utilize lead-acid (LA) battery.

Fig. 3
figure 3

HyPV D3 installed in a house (Taipei)

Full size image
Fig. 4
figure 4

HyPV D5 installed in an apartment (Taipei)

Full size image
Fig. 5
figure 5

HyPV D8 installed in laboratory (Taipei)

Full size image
Fig. 6
figure 6

HyPV D11 installed in a house (Chia-Yi)

Full size image

The microprocessor-based power control unit (MCU) monitors the instantaneous performance of HyPV. All the data are remotely transmitted to a PC in the laboratory at National Taiwan University, except D5 without internet.

3 Field test results

The four HyPV systems are run continuously on side.

3.1 HyPV-D3

Figure 7 shows the performance of thermal storage in D3. On 2016/4/19 and 2016/4/20, the PV power drives the water heater to store PV energy as heat since there is no electric load in daytime. The hot water heater stores PV energy from 42 to 72 °C, about 4.2 kWh. The water heater is boosted to 70 °C (set temperature) at sunset if the PV energy is insufficient for hot water supply at night.

Fig. 7
figure 7

Performance of solar PV thermal storage in D3

Full size image

Figure 8 shows the long-term performance of D3 with PV energy storage as heat. The hot water temperature reaches 70 °C in winter, spring, and fall when the air conditioners are not used. Figure 9 shows the daily energy supply from solar PV and grid. It shows that the solar PV supplies 24% of the total load. The long-term average specific PV energy generation of D3 in 2016 (Fig. 10) is 1.84 kWh/kWp per day (Fig. 9) which is about 20% lower than that of FIT solar PV system (2.47) in Taipei. This is due to the shadow effect caused by surrounding buildings which is quite serious in afternoon between October and March.

Fig. 8
figure 8

Performance of D3 with PV thermal storage

Full size image
Fig. 9
figure 9

Energy supply of D3

Full size image
Fig. 10
figure 10

Daily specific PV energy generation of D3 in 2016

Full size image

3.2 HyPV-D5

D5 was installed on the rooftop of an apartment building. No remote monitoring system was installed. The performance was monitored through regular metering. Table 2 shows the average performance of D5 in about a year. The average specific PV energy generation of D5 in 2016 is 2.59 kWh/kWp per day which is about 5% higher than that of FIT solar PV system (2.47) in Taipei.

Table 2 Average performance of D5 (2015/5/21-2016/5/31)
Full size table

3.3 HyPV-D8

D8 was installed in the laboratory with regular hot water drainage control to simulate real application in home. Figure 11 shows the water temperature variation during PV thermal storage. It is seen that solar PV energy was stored as heat when the battery is fully charged and the load is not in use. D8 also performed boost heating at sunset if the water temperature is below 70 °C. Table 3 is the daily performance of D8 in May 2018. Figure 12 shows the average specific PV energy generation of D8 in May 2018 is 4.05 kWh/kWp per day which is about 64% higher than that of FIT solar PV system (2.47) in Taipei. This is due to the good weather in May which is better than the whole year average.

Fig. 11
figure 11

PV thermal storage of D8

Full size image
Table 3 Daily performance of D8 in 2018
Full size table
Fig. 12
figure 12

Daily specific PV energy generation of D8 in May

Full size image

3.4 HyPV-D11

D11 was installed in early 2018 in a resident house in Chia-Yi (southern part of Taiwan). It is clearly seen from battery voltage variation that PV power was converted into heat storage when the battery is fully charged as shown in Figs. 13 and 14. Table 4 shows that the average specific PV energy generation of D11 is 4.35 kWh/kWp per day which is about 13% higher than that of FIT solar PV system (3.84) in Chia-Yi. This is because the data collected are not long enough to cover the whole year.

Fig. 13
figure 13

Performance of solar PV thermal storage (1/25)

Full size image
Fig. 14
figure 14

Performance of solar PV thermal storage (3/26)

Full size image
Table 4 Test results of D11
Full size table

3.5 Overall performance comparison

The performance of four HyPVs (D3, D5, D8, and D11) can be compared. It is seen from Table 5 that the average specific PV energy generation is all near that of FIT solar PV system installed locally, except D3 which has serious shadow effect between October and March. This verifies that the dual energy storage in HyPV can eliminate solar PV generation loss due to system mismatch. By using thermal storage, the cost of energy storage can be reduced.

Table 5 Overall performance comparison of D3, D5, D8, and D11
Full size table

4 PV production cost analysis

To analyze the PV production cost, a long-term system analysis is carried out using the performance model of HyPV and meteorological data with different system design parameters and load pattern.

4.1 System simulation model

In the HyPV, the incident solar radiation (IT) is absorbed and converted into electricity Ppv which drives the load (PL) and charges the battery (Pbat), as shown in Fig. 15. The energy balance equation, Eq. (1), is derived [1].

Fig. 15
figure 15

Energy balance of HyPV in PV mode

Full size image
$$P_{\text{bat}} = {\kern 1pt} {\kern 1pt} P_{\text{pv}} - P_{\text{L}}$$
(1)

The battery charging will be cut off and the PV power generation approximates the load power, if it is fully charged. This will results in a PV generation loss. That is, there are two operating states in PV power generation as described in Eq. (2).

$$\frac{{{\text{d}}E_{\text{b}} (t)}}{{{\text{d}}t}} \approx \left\{ {\begin{array}{*{20}l} {P_{\text{pv}} - P_{\text{L}} ,} \hfill & {{\text{battery}}\,{\text{not}}\,{\text{fully}}\,{\text{charged}}} \hfill \\ {0,} \hfill & {{\text{battery}}\,{\text{fully}}\,{\text{charged}}} \hfill \\ \end{array} } \right.$$
(2)

where Eb is the battery storage level (i.e., state of charge—SOC).

4.2 System analysis

The long-term system simulation for different design parameters was carried out using the local meteorological data. Four load patterns are used in the present study: load pattern A (100% energy used in daytime 6:00–18:00), load pattern B (75%), load pattern C (50%), and load pattern D (25%). Total PV energy generation over 20 years was calculated. The total investment cost includes initial and maintenance costs. Table 6 is the cost structure of HyPV solar home system used in the economic analysis.

Table 6 Cost of HyPV components used in economic analysis
Full size table

Three parameters are defined, Eq. (3), to correlate the long-term system performance results:

$$R_{pL} = \frac{{P_{pv} }}{{P_{L} }};\quad R_{hb} = \frac{{E_{h} }}{{E_{bat} }};\quad t_{bp} = \frac{{E_{bat} }}{{P_{pv} }}$$
(3)

RpL is defined as the ratio of maximum PV power generation (Ppv) to load power (PL); Rhb is the ratio of thermal storage (Eh) to battery storage (Ebat); and tbp is the charge time for solar PV to fully charge the battery.

4.3 PV energy production cost

Figure 16 is the variation of PV energy generation cost with RpL. It shows that there is an optimal design for PV generation cost. The optimal ratio of maximum PV generation Ppv to load power PL (RpL) is between 1.7 and 2.2.

Fig. 16
figure 16

Variation of PV generation cost with RpL

Full size image

Figure 17 is the variation of PV generation cost with Rhb. An optimal design for PV generation cost can be found. The optimal ratio of thermal to battery storage capacity (Rhb) is between 1 and 6, depending on tbp (ratio of battery size to PV module installed). It is seen from Figs. 16 and 17 that the PV energy production cost decreases with decreasing tbp. That is, smaller battery is more economic.

Fig. 17
figure 17

Variation of PV generation cost with Rhb

Full size image

The grid electricity price in Taiwan is categorized in staircase structure (higher electricity consumption paying higher unit price). The grid electricity price forecast for the next 20 years is shown in Fig. 18, according to the government policy. The PV generation cost from system simulation for 20 years at different designs is then compared with the average grid electricity price. Table 7 shows the optimal design for four kinds of load patterns in three cities of Taiwan. It is seen that the PV energy production cost is among 4.66–7.24 TWD/kWh (1 USD = 31 TWD). For load pattern A–C, the present HyPV solar home system with dual energy storage is economic if it was used to substitute partial energy demand (29–59%) which is paid at higher grid electricity price (Category 3 or 4). For load pattern D (worst case), the solar home system is economic if it is used to substitute 17–20% of daily energy demand which is paid at higher grid electricity price (Category 4 or 5). This is the partial grid parity.

Fig. 18
figure 18

Grid electricity price in Taiwan (1 USD = 31 TWD)

Full size image
Table 7 Optimal design and economic feasibility
Full size table

4.4 Cost reduction of dual energy storage

The above economic analysis is for HyPV solar home systems with Li battery and hot water storage. For the four HyPV solar home systems built and tested as described in Sects. 2 and 3, only D5 utilizes Li battery. We can compare the total energy storage cost at various combinations. Using the same design of HyPV D3, D5, D8, and D11, we can calculate the total energy storage cost at various combinations as shown in Table 8.

Table 8 Energy storage cost comparison
Full size table

It is seen that the dual energy storage would reduce the energy storage cost by 72–87% if using Li battery only and 41–52% if using LA battery only except D5. D5 utilizes a rather large heat storage (about 78% of total energy storage capacity), and this causes no cost reduction even using all LA battery to replace it. That is, using LA battery only  to replace Li battery and the cheaper and larger thermal storage (4.2 kWh) in D5 will increase the total energy storage cost because the heat storage used is relatively large. This can explain the existence of optimal RpL (ratio of maximum PV generation Ppv to load power PL) and Rhb (ratio of thermal storage Eh to battery storage Ebat) in Figs. 16 and 17.

5 Conclusion

In the present study, we developed a HyPV solar home system which generates solar power for self-consumption and utilizes dual energy storage. The solar home system utilizes a switching-type solar PV system (HyPV) which operates in either solar or grid mode automatically. No solar power is fed into grid. The combined battery and thermal storage can reduce the system cost since thermal storage is much cheaper than battery. An electric water heater is used to store solar PV energy when the battery is full. Four HyPV solar home systems were built for long-term field test. The test results show that the specific PV energy generation approximates that of FIT PV system.

An economic analysis was also carried out. The cost of dual energy storage as well as the system cost is reduced since thermal storage is much cheaper. It shows that the PV energy generation cost is lower than the electricity price from grid if the system is properly designed. That is, HyPV solar home system with dual storage is economic if it was used to substitute partial energy demand which is paid at higher grid electricity price. This verifies the possibility of grid parity for a solar PV system with energy storage.