Abstract A mobile-bed, undistorted physical model (1:40) has been used to investigate different sediment supply strategies to the Old Rhine through bank protection removal and modifications of groyne dimensions and configuration, which cause bank erosion. This trained channel was previously the main bed of the upper Rhine downstream of Basel (Switzerland), but it has an artificially low flow regime since the construction of the Grand Canal d'Alsace, a navigation canal, and a flow control dam at Kembs (France). Training works and subsequent channel incision have also greatly reduced sediment transport rates and created a heavily armoured bed. The modelled pilot site has a groyne field on the left bank. Results show that the currently existing groynes at the site are not effective in creating high bank-side velocities conducive to bank erosion, even for flow rates significantly higher than the mean annual flow rate. The river bank has also proved to be more resistant than previously thought, allowing long stretches of bank protection to be safely removed. The physical model testing process has produced a new configuration for the groyne field, where two higher, larger island groynes are placed further apart than the three existing attached groynes. This innovative approach has proved effective, causing bank erosion for flow rates below the mean annual flow rate, with consistent erosion being observed. It has also been found that such a configuration does not pose a hazard for the Grand Canal d'Alsace, which is situated next to the Old Rhine, through excessive bank erosion during high flow events.