Solar photovoltaic energy production relies on the direct conversion of sunlight into electricity. Silicon thin-film solar cells represent a reliable and environmentally friendly technology to make photovoltaics economically viable. As these solar cells do not contain any rare or toxic materials they are an ideal candidate for the large scale deployment of solar energy. At the same time photovoltaic energy production is ideal for off-grid, stand-alone installations. This thesis presents new methods and results on computational engineering of thin-film silicon solar cells and modules. Computational engineering employs numerical methods and simulation techniques to address engineering challenges. Predictive numerical simulations allow for device optimizations that otherwise require substantial experimental effort and they may lead to new insights into the device physics which is the basis for further improvements in terms of cell efficiency and lifetime. We discuss the development of two new simulation methods and present several case studies that illustrate their use in solar cell characterization and optimization. The first presented method addresses optical simulations, i.e. the interaction of the solar cell with sunlight. Our optical model is based on the net-radiation method that considers incoherent, ray-like propagation of light. We extend the net-radiation method with thin-film optics to allow for arbitrary sequences of thick and thin layers. Experimentally, advanced techniques such as optimized surface textures have been developed to increase the absorption in the solar cell and thereby to increase the produced current. We illustrate how our optical model takes into account these textures by means of azimuth integrated light scattering properties to formulate a computationally efficient model of light propagation within the solar cell. Our optical model thus integrates coherent and incoherent propagation of light as well as a non-iterative treatment of ligth scattering. It bridges the gap between wave and ray optics that alternative approaches are most often facing. We first apply the method to study a recently proposed rear electrode architecture based on lithium fluoride and aluminum in amorphous silicon solar cells. Our model-based optical analysis identifies the increased reflectivity of this back electrode as its main advantage. Secondly, we then explore the potential performance gains in microcrystalline silicon solar cells that can be reached by eliminating absorption in layers that do not contribute to the photogeneration of solar cell current. Finally, our optical model is also suited for other solar cell technologies and we illustrate its extension and application to light extraction calculations in light emitting diodes (LEDs). Our optical model has been incorporated into the software package setfos, commercialized by Fluxim Inc. It is thus available to the research community. We then turn to address large area solar modules. A new method based on the finite element method (FEM) is presented to calculate the electrical charge and heat transport in these devices and how it is influenced by defects as well as partial shading. We identify such defects with surface temperature measurements by means of lock-in thermography and demonstrate how our model accounts for the thermal signature of localized defects. Taking into account the large aspect ratio of thin-film solar modules we show how geometrical projections can be used to reduce the model complexity, leading to a numerically efficient approach based on the finite element method. We demonstrate this by presenting full current-voltage curves of solar modules under partial shading.