Geothermal energy potential is usually discussed in the context of conventional or engineered systems and at the scale of an individual reservoir. Whereas exploration for conventional reservoirs has been relatively easy, with expressions of resource found close to or even at the surface, exploration for non-conventional systems relies on temperature inherently increasing with depth and searching for favourable geological environments that maximise this increase. To utilitise the information we do have, we often assimilate available exploration data with models that capture the physics of the dominant underlying processes. Here, we discuss computational modelling approaches to exploration at a regional or crust scale, with application to geothermal reservoirs within basins or systems of basins. Target reservoirs have (at least) appropriate temperature, permeability and are at accessible depths. We discuss the software development approach that leads to effective use of the tool Underworld. We explore its role in the process of modelling, understanding computational error, importing and exporting geological knowledge as applied to the geological system underpinning the Guangdong Province, China.