Implementation of Numerical Techniques for Estimation and Investigation of Photovoltaic Heat Island (PVHI)

Rapid urbanization has had a large role in contributing to the Urban Heat Island (UHI) which translates to warmer air temperatures. Solar energy or photovoltaics (PV) is one of the most widely used renewable sources of electricity. With the introduction of PV systems in the urban landscape, little is known of the extent to which such systems contribute to the UHI. This study aims to investigate the Photovoltaics Heat Island (PVHI) effect through numerical analysis and derive a simplified model to estimate the temperature due to the PVHI. A numerical analysis was conducted to comprehend the extent of the PVHI on a 16kWp PV system in the SIT@Dover campus, located in Singapore. Three MX2302A temperature sensors were installed directly above three surfaces: PV system, rooftop, and grass field to capture temperature data at a logging interval of 5 min. A HOBO U30 USB Weather Station Data Logger in the vicinity was set at 5 min logging interval to record ambient temperature, irradiance and windspeed data. A clear day 24-h mean temperature data revealed that the PV system elevated the air temperature directly above it by 2.97 °C in comparison to the ambient temperature. On the same day with the overhead sun between 1200 – 1400hrs, the PVHI effects were accentuated 13°C above the ambient temperature. A linear regression technique known as Analysis of Variance (ANOVA) was implemented to derive a multi-variate linear equation to predict the temperature resulting from the PVHI. The statistical significance indicators of each of the three independent variables (ambient temperature, irradiance and windspeed) were p<0.05. The equation was tested and validated on seven days of actual data. The coefficient of determination of the calculated and actual PVHI temperature was found to be r² = 0.917, and the mean variation between the actual and calculated PVHI temperature was found to be 1.67°C, both indicative of a reliable model for estimation.