In 2015, the Dutch government initiated the ‘longitudinal training dams pilot project’. With the help of longitudinal dams, a two-channel system is created in which the river is divided into a main and a side channel. A combination of an inlet and several openings between the dams allows water and sediment to be exchanged between the two channels. This is expected to influence the discharge (and sediment) distribution between the main and side channel. Understanding the hydraulic conditions and the impact of the regulatory inlet and opening crest heights is expected to improve the control strategy and future design of river interventions using the concept of longitudinal training dams. A one-dimensional (Matlab) and a two-dimensional (Delft3D-Flow) model are created for this research. Both schematize the two channels with dams and openings. The influence of several mechanisms, factors and dimensional design parameters on the discharge distribution is assessed. Additionally, using Delft3D-Flow, the detailed flow patterns around the inlet and openings are evaluated and the schematizing of the inlet as part of the local bed topography or by using subgrid weirs is compared. From the several model runs, the discharge distribution shows to be mainly controlled by differences in longitudinal water level slopes in both channels, caused by the combination of too short inlet and opening lengths. It is not influenced significantly by the placement of the longitudinal training dams in a river bend. For increased crest heights the flow conditions change from submerged to free-flowing. In submerged conditions the velocity magnitudes and flow angles atop the inlet and opening crests are shown to increase in downstream direction while for free-flowing conditions they remain nearly constant. The free-flowing conditions are well described by empirical relationships, while for submerged conditions an additional angle submergence coefficient is suggested to parameterize flow angles. After some modifications, which will be included in the new Delft3D-Flow software release, schematization as part of the bed topography and subgrid weirs both predict similar discharges (< 7% relative difference). If interest is only given to water levels and discharge distributions, the use of subgrid weirs is recommended in Delft3D-Flow. For models assessing other river functions requiring local flow patterns or sediment transport, schematization as part of the bed topography is essential. However, by including the suggested parameterization for flow angles, the errors in flow angles for subgrid weir can be reduced significantly. From this research, the adjustable inlet and opening crest heights do not show a large influence on the discharge distribution (5-10% difference). It mostly impacts the tipping point for submerged to free-flow conditions. In submerged conditions, the inlet crest height does reduce the specific discharge in the side channel, limits the transverse velocities in the main channel, and greatly influence the flow angles atop the crest. Besides the crest heights, the channel width and inlet length should also be included as adjustable parameters in the design or control strategy of new systems of longitudinal training dams.

The one-dimensional model provides useful insight inm impacts of regulatory design parameters of longitudinal training dams. Preliminary conclusions include: 1. Adjusting the inlet weir height could influence the discharge distribution towards the side channel between 1% (fully closed but porous inlet) to 30% (fully open and porous inlet). 2. Smaller side channel width (e.g. sedimentation of the side channel) will result in significant reduction of discharge distribution toward the side channel. 3. The porous discharge through the inlet and openings is negligible compared to the weir discharge.