We propose a test procedure to quantify the response of dry sand to cyclic compressional loading under constrained conditions. The test procedure is designed to represent the response of sand layers to upward propagating P-waves during an earthquake event. Such P-waves are prominent within the vertical component of earthquake ground-motions, which is often ignored or simplified in common practice of seismic hazard analysis, despite its potential damaging effects. In the proposed method, the lateral deformation is restrained within a triaxial device, through variations of the cell pressure, thus maintaining pure compression while allowing moderate to large axial strains. Both dry and saturated samples are tested, and the compressive stiffness is computed from the full stress-strain loops. We show that as long as drained conditions are maintained and volume changes are allowed - the response of a saturated sample to slow cyclic loading is representative of the response of dry sand to seismic loading, despite the differences in saturation and in strain rates. Finally, we compare the proposed method to cyclic loading within a rigid cell and discuss the differences and limitations that the new proposed method overcomes.