Mixed order pressure-velocity approximations
 

I was wondering if anyone had any experience using mixed order pressure velocity formulations. Specifically, quad elements with bi-linear interpolation for velocity (stored at nodes) and piecewise constant for pressure (stored in the elements).

It seems to me that this is a convenient way to implement "staggered grids" on unstructured meshes, and I have always found that staggered grids are more stable than collocated (P and V at the same nodes) - at least for orthogonal meshes.