Modelling sediment resuspension in industrial tanks using SPH

Abstract

The present paper uses the weakly compressible smoothed particle hydrodynamics (SPH) to model numerically the sediment suspension and the shear layer induced by rapid flows resulting in scouring of the sediment with two yield criteria based on the Mohr-Coulomb and Drucker-Prager formulations. To the best of the authors’ knowledge this is the first time that the two yield criteria have been rigorously assessed in SPH where they are both expressed in terms of Coulomb-type parameters. The sediment is treated as a pseudo-Newtonian fluid where two different conditions distinguish between the state of the sediment – either still or suspended- according to the yield criteria. A variable viscosity is used to depict accurately the suspension layer. The models have been implemented using the DualSPHysics code, an SPH solver capable of dealing with high deformations and free surface flows. Three validation cases have been used to assess the validity of the yield criteria. The still sediment/water case was used to investigate the stability of the interface and the interaction with the boundaries that produced acceptable results for both criteria. The global relative error in the still case was used to assess the yield behaviour of the solid phase with both models performing satisfactorily. It was shown that the Mohr-Coulomb criterion was more conservative in particle movement. The second validation case was the tangential flow between two co-rotating co-axial cylinders using an analytical solution to confirm the rheological behaviour of the viscous sediment. With this case it was shown that the variable viscosity of the Drucker-Prager is lower than the Mohr-Coulomb model making the sediment less viscous. Finally, the erosional dam break was used to visualize the scour induced by the fluid. Following on from the results of the previous test cases in terms of variable viscosity and interface interaction the scour predicted with the Drucker-Prager model was larger and closer to the experimental results with a larger and deeper shear layer capturing the dune and scour. The Mohr-Coulomb yield criterion appears to underpredict the scouring process in addition to a small lag in the toe front of the solid phase that may be accounted to the smaller shear layer of the sediment. A sensitivity analysis on the Coulomb parameters and a convergence study must be performed in order to assess any deviations from the results discussed herein.