To enhance ultimate recovery of hydrocarbon gases from unconventional gas resources such as shales, we need to uncover the non-intuitive gas transport mechanisms in ultra-tight porous media. Exploiting our previous and recent pioneering work in modelling rarefied gas flows at micro/nano-scales and in pore-scale characterisation of reservoir rocks, we present an ambitious project to tackle this newly-emerged research challenge through developing direct numerical simulation models and techniques that work on binarised images of concerned porous materials. This work will transform the currently-adopted heuristic approaches, i.e. Darcy-like laws and pore network modelling, into those underpinned by the first principle, and enable the quantification of prediction uncertainty on gas transport associated with the former approaches. Timely support now from EPSRC will provide us crucial resources to shape this emerging research area - understanding and quantifying gas flow physics in ultra-tight porous media.