Finding a generalized compaction model for 19 oedometric tests on granular salt
The time required to compact initially loose granular salt into an impermeable backfill body, for example for the final disposal of radioactive waste in salt formations, is foremost determined by the on-site rate of cavity convergence. However, this rate decreases with the increasing load resistance of the salt backfill, which, in turn, is controlled by a range of complex interacting factors, whose relative contributions have not yet been fully quantified. Hence, forecasting when a backfill body will reach its impermeable compaction state is still challenging. In this empirical approach, we re-evaluate 19 oedometric compaction tests, in which the increase of backfill resistance was experimentally determined for varying boundary conditions. We show to what extent the increase varies across all experiments and highlight how this variation correlates with the deliberately varied test conditions. To this end, polynomial prediction models of varying degrees were fitted for each test. Based on the adjusted coefficient of determination $$ \left({R}_{adj}^2\right) $$, a fourth-degree polynomial was found to provide the best representation of the observed variability within the data $$ \left({R}_{adj}^2=0.27\right) $$. When fitting prediction models based on data grouped by compaction rate the mean adjusted R 2 improves over all groups to 0.67 leading to a more precise estimation of the underlying data and provides a promising outlook to eventually extrapolate for in-situ compaction rates. Further data grouping seems valuable if additional tests (1) allow a well-maintained moisture content throughout the test duration and (2) expand the value range and distribution for temperature, grain size and material type.