Opto-Geometric Modelling of Complex Urban Landscapes

Abstract

The injection of renewable energy into the grid requires meticulous planning and execution due to the unpredictable nature of weather and in-turn the energy generated by these sources. Moreover, the inclusion of grid-independent energy generators requires accurate tools to predict energy yield in order to simplify matters such as grid-congestion management. One such tool useful to compute solar energy yield, which is under development, is the PVMD toolbox from the PhotoVoltaic Devices and Materials research group within the Delft University of Technology. This toolbox, which is completely based on MATLAB , allows users to specify parameters ranging from the micro-scale (such as the chemistry and characteristics of a single cell) to the macro-scale (such as the electrical parameters of the whole system).
However, in its present version, the toolbox can simulate the yield only for
simple systems, such as a single PV panel without any obstructions present nearby.
This makes it incompatible to solve complex problems such as models with solar panels installed in the built environment.

Therefore, the primary focus of this thesis is on the optical side of the PV simulation, by enabling the toolbox to precisely compute reflected component of irradiance and include shading effects on modules from nearby objects in the scene. This functionality will help consider the presence of features such as dormers, chimneys and highly reflective surfaces present in urban environments which contribute to solar power output. To achieve this, attention was paid towards the formats and sources of geometric data (3D objects in particular) which could be used as an input in the toolbox. Once the data format was selected, the 3D object is parsed into a MATLAB readable format and refined in order to preserve geometric information.
Afterwards, the scene was completed by enumerating the geometric data of
the solar cells and placing them beside the object or on an external surface of the object. Optical properties were then obtained in the form of spectral reflectance data from NASA’ ECOSTRESS Library and converted into the appropriate format to assign these material properties to the geometries. This Opto-geometric data is then parsed back into a file which can be read by RADIANCE, a visual rendering software capable of backward ray-tracing. This ray-tracing technique is used to obtain the irradiance available for each solar cell at a given time instant.

The developed workflow is validated by comparing it with real data stored at the PV Monitoring Station within the TU Delft Campus. For the given data-set, the mean percentage error is computed to be 1.89% , with a standard deviation of 12.49%. This proves that the proposed workflow is suitable to visualise shading performance and the effect of reflected irradiance. This model can be further improved by considering parameters such as spectrally responsive albedo and inclusion of other geometry formats. The outputs from this project can also be used as inputs to compute the spectral absorptance of different layers within a solar cell, such as tandem solar cells or develop sensitivity maps for different modules present in the scene.