Even when a quantum algorithm has a favourable query complexity, the cost of the data-loading oracle can break quantum advantage. This talk will focus on block encodings, a popular data-input model, for unstructured meshes. Unstructured meshes are frequently used to solve computational fluid dynamics problems in complex geometries. Their lack of structure makes efficient block encoding of the corresponding matrices difficult. After introducing block encodings and cost metrics that must be considered, I will present a new method for arbitrary sparse-matrix encodings and put it into context with existing methods.