Uncertainty-aware sample mass determination for particle size analyses of gravel-dominated soil

Georg H. Erharter , Santiago Quinteros, Diana Cordeiro, Matthias Rebhan, Franz Tschuchnigg

Determining particle size distributions (PSD) of soils is a basic first step in many geotechnical analyses and guidance is given in different national standards. For ambiguous reasons, the recommended minimum sample mass (m_min) for the PSD-analyses of soils with a main component of gravel or greater is based on equations including the soil's maximum grain diameter (D_max). We claim that the recommended m_min is overestimated as D_max does not represent the relevant large soil fraction but only the PSD's uppermost outlier. Furthermore, the recommended m_min is not based on a specific sampling confidence (i.e. how closely does the sample’s PSD need to approximate the soil’s PSD?) and thus it is not clear why the m_min should even be necessary. We conducted Monte-Carlo simulation-based sieve analyses of soils consisting of gravels and cobbles and developed a new, practically applicable framework to determine m_min based on D_90 that also includes explicit consideration of sampling confidence. A survey was conducted that shows that there is no significant difference in how well operators are able to assess parameters like D_90 or D_max. Real sieve tests performed on three different soils corroborate the theoretical results and show that substantially lower sample masses yield PSDs with only marginal differences to PSDs from samples according to the standards. While the results are promising, they open up for new research questions about which geotechnical application requires which soil sampling confidence.