The thermal performance of solar collectors has a significant effect on the thermal efficiency of a SWH. This study uses glazed, metallic flat-plate solar collectors (DYY-B1, DYY Solar Industrial Co. Ltd.). The value of A_sc for each solar collector is approximately 1.97 m². The 10-12 risers are made of copper and the thickness of the tempered glass is 3 mm.

In Taiwan, a national standard (the Chinese National Standards, CNS 15165-1-K8031-1) is in compliance with ISO 9806:1994. It specifies an outdoor test method to determine the steady-state thermal performance of solar collectors (natural solar irradiance ≥ 800 W/m²). Chung et al. 23 also showed that the effect of diffuse solar radiation on the thermal performance of a glazed, metallic, flat-plate solar collectors is not significant. The test facility for the thermal efficiency of a solar collector with water load at the Energy Research Center, National Cheng Kung University (ERC/NCKU) is shown in Figure 1.

The solar collectors for this study faced south at a tilt angle of 23° to optimize the collection of solar radiation. A precision spectral pyranometer (Eppley, Model PSP; uncertainty = ±0.91%)was used todetermine the value of I. The water mass flow rate ṁ is 0.02 kg/s/m² and is determined using a flow meter (Macnaught, Model G2SSP-1R; uncertainty = ±0.1%) that is positioned along the water supply line. Three platinum resistance thermometers (Thermoway, Model PT100; uncertainty = ±0.1°C) were installed to record the ambient temperature, T_a, the initial temperature in the cold-water supply line T_in and the final outlet temperature in the solar collectors T_out. The wind speed was measured using an anemometer (Young Co., Model 05103L; uncertainty = ±0.03 m/s). Each test sequence used four input temperatures (T_a±3°C, 45°C, 60°C and 70°C). Data from the monitoring devices was sampled every second using a data acquisition system (M-7015p and M-7017z, Pactech, Inc.). The value of η_sc is calculated using Equation 1, and the calibration curve is shown in Figure 2. The intercept F_R(ta) is the useful energy that is obtained from a solar collector, and the gradient F_RU_L represents heat loss. The respective values are 0.776 and 4.66.

(1)

The feed plant is located in Pingtung County, Taiwan (22°22’11’’N, 120°35'10’’E; which is south of the Tropic of Cancer). It produces 0.2 million tons of soybeans, corn, and wheat per annum. Four boilers supply steam at 6 bar and 158°C. The daily water consumption is 30-40 tons. Heat is required for three stages: ambient temperature to 100°C (liquid, specific heat capacity = 4.2 kJ/(kg·K)), gasification (latent heat, 2257 kJ/kg), and steam superheating (100°C to 158°C, 498 kJ/kg). The plant consumes approximately 70,000 liters of low-sulfur light fuel oil per month. The price of low sulfur-light fuel oil varied between 400 and 854 US$/kL from 2020 to 2025 24.

A SWH with flat-plate solar collectors can supply hot water of 70°C. The monthly global solar radiation, G_m, as determined using the Kriging method 25, as shown in Figure 3. The value ranges from 310 MJ/m² (January-February) to 532 MJ/m² (July). This shows that the location of the feed plant is ideal for the use of solar thermal heating. A south-facing SWH was installed on a roof with a 6° pitch on a low-rise building to minimize shading, as shown in Figure 4. The area of the 60 glazed flat-plate solar collectors is approximately 118 m². Garg 26 showed that a parallel arrangement produces maximum thermal efficiency. This study uses three solar collectors in series (an array) with one inlet and one outlet to heat cold water. The arrays are then connected in parallel, as shown in Figure 5.

A hot water tank stores 2 tons of water. A reverse-type thermostat is used to control a circulation pump, which is switched on and off depending on the temperature difference (6°C/2°C for the baseline case) between the water storage tank and the water outlet for the solar collectors. The flow rate is 200 liters/minute. The temperature difference at which the thermostat is set has a significant effect on the system performance of a SWH. This study uses a programmable logic controller (PLC) to control the settings for the thermostat. The circulation pump is switched on and off (6°C/2°C; 8°C/2°C) if the value of I is greater than 400 W/m². For a low value of I (< 400 W/m²), a setting of 3°C/1°C aims to produce better performance for the SWH than the baseline case.

Several monitoring devices recorded the operating conditions for the SWH. A precision spectral pyranometer (Kipp & Zonen Inc., model CMP11)was used todetermine the value of I. A Macnaught flow meter (Model M2SSP-1R) was located in the cold-water supply line to the water storage tank to record hot water consumption, ṁ, and another flow meter was installed in the circulation line from the bottom of the water storage tank to the inlet of the solar collectors (mass flow = 200 LPM). Nineteen platinum resistance thermometers (Izuder Enterprise, 1/10 DIN Class B) measured the local water temperature and ambient temperature.

The data from the monitoring devices was sampled using ICP DAS analog input modules (Model M-7017; sampling rate = 1 Hz) and transmitted to the host computer at the ERC/NCKU via the Internet. The thermal efficiency of the system, η, is calculated using Equation 2:

(2)

C_p: specific heat; MJ/(kg°C)

ṁ: water mass flow rate; kg/s

T_out: final outlet temperature in the solar collectors; °C

T_in: initial temperature in the cold-water supply line; °C

The supply of Taiwan’s energy depends exclusively on imported fossil fuels. For the development of clean energy, the “Framework for a Sustainable Energy Policy" was announced in 2008. The “Renewable Energy Development Bill” and the “Greenhouse Gas Emission Reduction and Management Act (GGEMRA)” were also respectively enacted in 2010 and 2015. The “Climate Change Response Act, CCRA”, which originally read the GGEMRA, was amended and enacted in 2023. It defines Taiwan's net zero emissions for 2050, and the five-year periodic regulatory goals to attain zero emissions. The proposed targets are a 32±2% carbon reduction by 2032 and a 38±2% carbon reduction by 2035, compared to the baseline year (2005) 27.

Carbon taxes are a central climate policy instrument to decrease carbon emissions. Gielen 28 showed that the lowest carbon tax has a relevant emission-reducing effect of US$25–US$40 per ton of carbon dioxide (CO₂) equivalent tCO_2e for high-income countries and US$10/tCO₂ for low- and medium-income countries. Twenty-five national carbon taxes have been implemented during the period of 1991-2022 and the highest national carbon tax is 137 US$/tCO_2e in Uruguay.

Article 28 of the CCRA details the long-term goal for the reduction of national emission and periodic regulatory goals. The central competent authority can impose carbon fees in stages against direct and indirect emission sources. It stipulates the legal foundation for levying carbon fees. Carbon tax (300NTD/tCO_2e for the general rate; 1US$ 31.5NTD) was implemented in 2025. All manufacturing and power industries for which annual carbon emissions exceed 25,000 tCO_2e are billed. It is expected to strengthen carbon reduction targets 27. However, Lilliestam et al. 29 showed that the reduction in carbon emissions depends on the cost of the carbon tax. Low carbon taxes may only meet international expectations and do not always reduce carbon emissions.

The field measurements were conducted from February to May. The monthly duration of sunshine was 252-285 hours. As shown in Figure 3, the value of G_m ranges from 310-497 MJ/m2. Figure 6 shows the daily’s system performance for the baseline case (6°C/2°C) which is a typical setup for a forced-circulation SWH in Taiwan. The horizontal axis shows the daily solar radiation, G_d. The data shows that there is an increase in the value of η (= 0.488-0.530) as G_d increases, particularly for G_d ≥ 17 MJ/m2 (η > 0.5). This result is in agreement with that for the study by Lin et al. 20.

A lower value of G_d results in a decrease in the system’s efficiency, so tests for this study used a two-stage setting of 6°C/2°C (I > 400 W/m²) and 3°C/1°C (I < 400 W/m²). The system’s performance is plotted versus G_d in Figure 7. The value of η ranges from 0.501 to 0.571 and the average value is 0.545. This shows that a two-stage setup for the thermostat increases the system’s performance. More solar energy is collected if the value of G_d is less. Figure 7 also shows that the value of η (= 0.501-0.508) decreases for three days because there is a decrease in hot water consumption and a change in the operation of the feed plant.

Another two-stage setting of 8°C/2°C and 3°C/1°C is used to the control circulation pump. An increase in the temperature difference (8°C/2°C and 6°C/2°C) produces a decrease in electrical consumption because of the circulation pump. Figure 8 shows the relationship between the solar heat that is collected and the value of G_d. The cross-correlation between the two time series (G_dand solar heat) is determined. The correlation coefficient, g, is defined as Equation 3:

(3)

where N is the number of tests, V1’ and V2’ are the fluctuating component for the two time series, and s_V1 and s_V2 are the values for standard deviation. The value of γ is 0.97. This shows that solar heat is closely correlated with the value of G_d. The average value of η is 0.539, which is slightly less than the value for a two-stage setting of 6°C/2°C; 3°C/1°C. The variation in the value of η with G_d is shown in Figure 9. The data is scattered but there is an increase in the value of η as G_d increases.

Figure 10 shows the data from the field measurements for the baseline case. The efficiency of the SWH depends on the incident solar radiation, the thermal performance of solar collectors, and the hot water consumption pattern. As a pre-heating system (indirect mode), it produces 3.8%-4.3% of the heat demand (solar fraction) for the feed plant per month (steam at 6 bar and 158°C) which saves energy and reduces carbon emissions.

The capital cost for the SWH (A_sc = 118 m²) is approximately 43,000 US$. The plant consumes approximately 70,000 liters of low sulfur light fuel oil per month. A reduction in the price of low-sulfur light fuel oil creates a longer payback period. For low-sulfur light fuel oil at a price of 625 US$/kL and annual consumption of 840 kL, a supply of 4% heat demand produces annual fuel savings of approximately US$21,000, so the simple payback period is 2.05 years. This is much less than the expected service period of 15 years for a SWH 20. This results shows that it is financially viable to use a SWH for the feed plant and an industrial heat process.

Sermab et al. 30 showed that low-sulfur light fuel oil has a carbon emission factor of approximately 3.114 kg/L (or 11.788 kg/gallon). The feed plant consumes approximately 70,000 liters per month so carbon emissions per annum are 2616 tCO_2e. Supplying 4% heat demand corresponds to a reduction in carbon emissions of 105 tCO_2e. Therefore, the annual carbon emissions for this feed plant are not within the scope of the carbon tax program in Taiwan (≥ 25,000 tCO_2e). The results of this study show that the SWH for this feed plant is financially viable, but there is no add-value (carbon tax as additional operating fee). In term of these results, the low carbon tax scheme and annual carbon emissions criterion for Taiwan cannot be used in isolation as a climate policy instrument. A revised scheme (a lower criterion for levying carbon fee or higher carbon tax) is required for the net-zero transformation process.