Bias due to side wall friction in sand box experiments

The principal concern of this paper is with the accuracy and validity of experimental sand box models of geological structures – and of thrust faults in particular – as influenced by side wall friction. Side wall friction can produce undesirable artifacts, in particular in experiments aimed at reproducing plane deformations. The importance of such artifacts is revealed in the present study by the results of experiments that are affected in opposite ways by side wall friction. Two distinct experimental configurations have been chosen for this purpose for a set-up that comprises a body of sand placed between vertical side walls on a basal plate against a back wall and a front wall, all made of glass. In configuration M the bottom plate and attached front wall move relative to the side walls towards a fixed back wall, while in configuration F the back wall moves relative to the side walls and attached bottom plate towards a fixed front wall (‘fixed’ = ‘fixed relative to the side walls’). In both configurations, thrust traces on the top surface of the sand become curved near the side walls because of wall friction. The outcome of altogether 31 experiments in the two configurations shows that the magnitude of the experimental bias depends on the ratio SL/SB of the area of contact of the sand body with the side walls (SL) to its area of contact with the basal surface (SB). The bias is the difference in thrust locations in the two configurations. It is measured on photographs taken through the side walls during the experiments. Away from the side walls, the bias is estimated by the thrust curvatures observed in top views. For SL/SB ≤ 0.1, i.e. very little contact with side walls, there is no identifiable bias. Furthermore, a 2D theoretical solution reproduces well the results. For SL/SB increasing to 0.25, curvatures increase, yielding rather different cross-sections at the side walls, but still similar cross-sections in the central part of the sand body. For SL/SB ≥ 0.3, thrust planes occupy different regions of the box for each configuration, although their curvature is decreasing. At SL/SB = 0.9, curvatures have almost disappeared but the location of the thrusts and the length of the décollement depend strongly on the experimental configuration. Departures from the 2D theoretical solution are clearly stronger with configuration M than F. Another important conclusion is that the absence of curvature of the thrust traces is not a proof of the 2D character of the experiment. Mechanical equilibrium of the total forces recorded at both front and back walls during shortening shows that the thrust locations are controlled by the difference between basal and side wall shear forces.

► In analogue models of 2D tectonics, the side walls exert spurious shear forces.
► These forces can create experimental biases (curved and/or misplaced thrusts).
► The bias is stronger in narrow boxes using a basal conveyor belt.
► Planarity of thrust surfaces does not imply the absence of bias.
► Difference between basal and side shear forces control thrust locations.