Benchmarking analogue models of brittle thrust wedges

International audience ; We performed a quantitative comparison of brittle thrust wedge experiments to evaluate the variabilityamong analogue models and to appraise the reproducibility and limits of model interpretation. Fifteenanalogue modeling laboratories participated in this benchmark initiative. Each laboratory received ashipment of the same type of quartz and corundum sand and all laboratories adhered to a stringentmodel building protocol and used the same type of foil to cover base and sidewalls of the sandbox. Sievestructure, sifting height, filling rate, and details on off-scraping of excess sand followed prescribedprocedures.Our analogue benchmark shows that even for simple plane-strain experiments with prescribedstringent model construction techniques, quantitative model results show variability, most notably forsurface slope, thrust spacing and number of forward and backthrusts. One of the sources of the variabilityin model results is related to slight variations in how sand is deposited in the sandbox. Small changes insifting height, sifting rate, and scraping will result in slightly heterogeneous material bulk densities,which will affect the mechanical properties of the sand, and will result in lateral and vertical differencesin peak and boundary friction angles, as well as cohesion values once the model is constructed. Initialvariations in basal friction are inferred to play the most important role in causing model variability.Our comparison shows that the human factor plays a decisive role, and even when one modeler repeatsthe same experiment, quantitative model results still show variability. Our observations highlightthe limits of up-scaling quantitative analogue model results to nature or for making comparisons withnumerical models. The frictional behavior of sand is highly sensitive to small variations in material stateor experimental set-up, and hence, it will remain difficult to scale quantitative results such as number ofthrusts, thrust spacing, and pop-up width from model to nature.