Dynamic photovoltaic (PV) greenhouses integrate sustainable energy generation with plant cultivation, offering more possibilities of energy production and microclimate control by adjusting the sun-tracking angles. Previous studies on PV greenhouses barely paid attention to the PV partial shading effects, and rarely recorded the performance across the full range of rotation angles. In this study, we first build computer simulation models of typical greenhouses with high-density (1/2 roof area) and low-density (1/3 and 1/4 roof area) PV layouts. Then four special sun-tracking positions are found in the model of equivalent global irradiance, which is defined as the quotient of the total input power divided by the area of PV module under partial diffuse shadows. Simulation models are also built in terms of PV modules and interior irradiance. Simulations are conducted using the climate data of Delft, the Netherlands (52.01°N,4.36°E). Results show that high-density PVs under no-shading sun tracking generate 6.91% more energy than that under conventional (quasi-perpendicular) sun-tracking. Meanwhile, no-shading sun tracking allows more diffuse sunlight to enter the greenhouse mounted with high-density PV panels, resulting in 10.96% and 10.68% improvement on the annual average global irradiance and uniformity on the target plane compared to the fixed PV panels in the closed position. Regarding low-density PV layouts, which barely suffer from partial shading problems, quasi-perpendicular sun tracking improves the annual energy generation by 7.40% relative to the closed position. However, the average global irradiance reaches the minimum in this position because more sunlight is blocked by PVs. Meanwhile, the average uniformity of global irradiance reveals good (but not the best) performance, resulting in up to 9.80% (1/3 coverage) and 4.70% (1/4 coverage) improvement respectively compared to the closed position. The proposed methods and simulation results provide guidelines for the initial design and daily operation of PV greenhouses, aiming to balance the PV power generation and food production.

@en

Current photovoltaic (PV) industrial chain mainly serves the conventional utility-scale PV power stations. Rigid and opaque silicon-based PV modules domain the market so far. As distributed PV capacities expand, PV modules tend to be integrated with existing infrastructures (mostly, buildings). To adapt with the building environment, innovative design is required from the cell level to the system level. This dissertation specially deals with the window-integrated photovoltaics. Two types of PV windows, those with opaque PV shading elements and those with semi-transparent PV (STPV) glazing, are mainly explored in terms of concerned performances. A mathematical model of solar irradiance and a geometrical model of a reference office are built in Chapter 2. One-axis PV blinds and the total input power are modeled and analyzed in regard to annual power generation and glare protection. An optimal sun-tracking angle has been found to achieve both maximum power generation and non-glare daylighting. Optimal design of cell layout is also proposed to avoid shading from window frames. Compared with conventional quasi-perpendicular sun tracking, the proposed sun-tracking methods improve the annual energy generation by 12.00% and the annual average efficiency by 8.52%. In Chapter 3, PV shading elements with extra degree of freedoms (DOFs) have been modeled and analyzed in a similar way as in Chapter 2. Two-DOF PV shading elements have been proved to be the same as one-axis PV blinds in respect to optimal sun-tracking positions. PV shading elements with three-DOF sun-tracking abilities are demonstrated capable to meet all the requirements, i.e. gaining the maximum power generation, protecting from glare, and avoiding shadows from the window frame. A corresponding variable-pivot three DOF (VP-3-DOF) sun-tracking algorithm is given in the form of an analytical solution. Following aforementioned two chapters, the overall energy performance of the reference office with one-axis PV blinds is analyzed over an entire year in Chapter 4. Simulations show that using the optimal shade-free tracking method, the net energy consumption of the building, considering PV production, artificial lighting, heating and cooling, is reduced by 10.49%, compared to the perpendicular tracking method. In Chapter 5, PV windows are applied to the skylight in Dutch greenhouses. Unlike vertically-mounted PV windows mentioned above, the greenhouse PV panels are installed on a pitched roof to regulate the sunlight for plants, instead of humankind. PV layouts in high and low densities are evaluated under four special sun-tracking positions with regard to power generation and interior irradiance. Simulation results provide guidelines to balance the PV power generation and food production in greenhouses. In Chapter 6, semi-transparent thin-film amorphous silicon solar cells are designed and fabricated for PV windows. Using an optical model, GenPro4, we provide with a simulation method to optimize the configuration of such solar cells. According to the optimized results, we fabricate the single-junction amorphous silicon solar cell, showing an average transmittance of 20.04% with the conversion efficiency of 6.94%. Additionally attached to a polymer dispersed liquid crystal (PDLC) film, which can switch from opaque to transparent state in a second by applying an alternating-current (AC) voltage, the transmittance of the PV window can be further controlled. The prototype of a house model, containing the STPV-PDLC system, has been built to demonstrate the feasibility of such a combination. Besides building applications, PV windows can be further applied to any occasion that requires light transmittance and power supply, such as electric vehicles, aircrafts, billboards, and even mobile phones. Applications of PV windows could be beyond imagination. @en

Vertical space bears great potential of solar energy especially for congested urban areas, where photovoltaic (PV) windows in high-rise buildings can contribute to both power generation and daylight harvest. Previous studies on sun-tracking PV windows strayed into the trade-off between tracking performance and mutual shading, failing to achieve the maximum energy generation. Here we first build integrated models which couple the performance of sun-tracking PV windows to the rotation angles. Secondly, one-degree-of-freedom (DOF) and two-DOF sun tracking are mathematically proven to be not able to gain either maximum power generation or non-glare daylighting under reasonable assumptions. Then we derive the optimum rotation angles of the variable-pivot-three-degree-of-freedom (VP-3-DOF) sun-tracking elements and demonstrate that the optimum VP-3-DOF sun tracking can achieve the aforementioned goals. When the restriction of the proposed model is relaxed, the same performance can be achieved by the optimum one-DOF sun tracking with extended PV slats and particular design of cell layout, requiring less complicated mechanical structures. Simulation results of nine global cities show that the annual energy generation and average module efficiency are improved respectively by 27.40% and 19.17% via the optimum VP-3-DOF sun tracking over the conventional perpendicular sun tracking. The proposed optimum sun-tracking methods also reveal better protection against sun glare. The optimum VP-3-DOF sun tracking is also demonstrated to be applicable to horizontal PV windows, as those applied in the sun roof of a glass greenhouse.@en

Diverse solar irradiance spectra can be observed under different conditions of time, date, location, weather, etc. Since the solar irradiance spectrum is required by certain scientific and engineering applications, obtaining accurate spectral data is essential. Measurements by spectrophotometers are able to achieve accurate real-time data with high resolution, but at high expense. While in some engineering applications, the requirements on accuracy and resolution are much lower than that in a typical scientific research. Therefore, a rapid method of estimating the solar spectrum is proposed based on an available spectral model in this paper. In order to achieve fast estimation, we simplify the input parameters of this model into five key inputs, including latitude and longitude, altitude, date and time, sky and ground type. The first three parameters are easy to obtain from GPS and the internet. Sky and ground types include common types of sky and ground, which can be input manually or processed automatically by analyzing a digital image of target sky or ground. The automatic input is realized through dominant color extraction or by training an artificial neural network. Results show that the proposed rapid method can generate different spectral power distributions based on distinct input conditions. Two device frameworks are also proposed to implement the rapid method, which is applicable to many fields. LED lighting is one of the most prominent applications. Users can easily share local sunlight with each other through an APP in mobile phones@en