Hi Luke,
Yep, you are on the track. The Boundary traction model can be used to prescribe tractions on boundaries, rather than velocities.
In general, models with both a free surface and boundary tractions on one or more side walls tend to be less stable than equivalent models with kinematic boundaries, but with enough testing various configurations you probably can arrive with a stable suite of simulations.
As lithostatic pressure will dominant for a crustal- or lithospheric-scale model, what you likely need to prescribe on the traction boundaries is the lithostatic pressure plus a constant that defines either inward or outward flow (simple 2D example).
Previously, I believe this combination could only have been done via a carefully designed user-specified function, but now it should be possible to combine multiple tractions models with the following recently merged PR (yay hackathons 
) - Problem with update strain rheology - #13 by jbnaliboff
As a starting point, I suggest reading through that PR (particularly the new example test case) and:
- Going through the online documentation for the boundary traction model
- Reading through relevant tests cases, including box_lithostatic_pressure_2D.prm. I would search for keywords like
lithostatic_pressureandboundary_traction. - Go through the benchmark suite for nonlinear channel flow, which uses boundary tractions to drive deformation.
I hope this helps and it would be great to continue the discussion with any further questions about existing examples or new tests you conduct.
Cheers,
John