Analysis of Solar Radiation and Daylight on Building Envelope and Implication to Building Energy Saving Designs


Student thesis: Doctoral Thesis

View graph of relations



Awarding Institution
Award date11 Aug 2017


In subtropical regions, the non-domestic buildings consume much energy because of the high solar radiation and the resulted air-conditioning load. In Hong Kong, commercial building consumes around 65% of the overall energy, and air-conditioning accounts for 31% of the building electricity uses. Meanwhile, building integrated photovoltaic (PV) can make use of solar energy, and the daylight linked controls can reduce the artificial lighting demands. Thus, the solar irradiance and daylight illuminance, and their impacts on the building envelope are essential to the building energy saving designs. This thesis studies the solar irradiance and daylight illuminance databases, estimation approaches, and discussed the implications to energy saving building designs.

The thesis uses 1) the hourly weather data from the Hong Kong Observatory between 2008 and 2015, including solar irradiance, air temperature, humidity, cloud cover, visibility; 2) the solar irradiance, daylight illuminance, and sky luminance data recorded in the City University of Hong Kong from 2004 to 2005 and in 2015; 3) and PV efficiency data from two local projects between 2010 to 2015.

The first part discusses the solar irradiance database development. The HKO has systematically recorded the global solar radiation on a horizontal plane since 1978. There are, however, no measurements of the diffuse and direct components until 2008. Besides, the sky luminance and radiance, and the vertical and inclined components are not recorded. The research firstly includes estimating the horizontal sky-diffuse irradiance and analysis its relevance to the climatic indices using a machine learning algorithm. It was found that the clearness index (i.e. ratio between the horizontal global and extra atmospheric solar irradiance), solar altitude, air temperature, cloud cover, and visibility were important parameters for evaluating the diffuse irradiance. The mean absolute error (MAE) of the logistic regression was less than 21.5W/m2 and 30 W/m2 for Hong Kong and Denver, USA, respectively. Then an approach was proposed to interpreting the CIE Standard Skies using the meteorological data which are readily accessible for many years. The approach properly identifies 83.2% of the three typical overcast, partly cloudy and clear skies, and further 62.7% of the 15 individual CIE Standard Skies for Hong Kong. The classified sky types can interpret the sky diffuse luminance and radiance from various directions, which affect the irradiance and illuminance on building envelopes. The root mean square errors of the vertical solar irradiance and daylight illuminance estimated by the numerical integration of the luminance and radiance were less than 23%.

The second part develops a model to estimating the solar irradiance on vertical and inclined building envelopes. Since the numerical integration of the irradiance and illuminance were complex, an approach was developed to simplify the diffuse solar irradiance estimation on the building envelope facing any directions. The ratios between the diffuse irradiance on a vertical, inclined and obstructed plane to that on the horizontal plane were defined as the Vertical, Inclined and Obstructed Sky Component (i.e. VSC, ISC, and OVSC), respectively. An approach to estimating such Sky Components under different CIE Standard Skies was established according to the building envelope direction concerning the sun. For the obstructed environment, a manual calculation procedure was developed to estimate the direct, diffuse and reflected solar irradiance on the vertical building envelope facades. Results were benchmarked with computer simulation and validated by field measurements in 2015 and 2016.

The third part is about the irradiance estimation model application. The energy performance for PV cells was studied to apply the solar irradiance estimations in building integrated PV designs. However, the PV nameplate performance relies on the laboratory settings which may rarely occur in practice. Thus, an artificial neural network (ANN) was developed to estimate the PV efficiency and energy production according to the weather data. The model predicts the year-round PV power production for different PV installations. The results are essential to the low and zero energy building designs if the energy demand is compensated by the PV cells.