As promised, here is the procedure I used. Use with caution as this probably will not work for all scenarios. It is a good step since the documentation on this is quite lacking.
Procedures for getting sediment transport to work:
1. Make sure the mesh is very stable when running hydrodynamics. i.e. its important to have as small of an elevation change as possible across elements. In addition, elements near the location of boundary conditions, especially the upstream one, are prone to cause instabilities. Here, it’s a good idea to use as small elements as possible with a maximum of 1-foot in elevation change across each element. Also, element area changes greater than 2 or 0.5 can cause instabilities elsewhere. This is not as vital as boundary elements, but it helps stabilize the model. If the steering module reaches a solution in 5 or less runs, then the mesh is ready for testing the application of sediment transport. In addition, if the velocities are greater than 15 fps anywhere, your model is not stable enough. Mine had a maximum velocity of about 7 fps, but I imagine larger velocities up to supercritical flow will also work.
2. With a strong mesh, enter your bed control parameters in the model control sediment control tab for feswms. Here enter the particle distributions with the sum of each percentage equaling one. Use 1-8 particle sizes. More particle sizes does not create instabilities, it only increases computational time. Also, active bed layer, deposition depth, and original depth are very important to stable sediment runs. The active bed layer should be approximately 2 times the median particle size (50-percent finer) of the study reach. Deposition layer of 1-3 feet was stable for all of my runs. The original bed layer is only stable if the depth is equal to or greater than the bed elevation of your upstream boundary condition. For example, if your upstream bed elevation is 600 feet, use 601 feet for the bed thickness. Values lower than this will cause your model to fail.
3. Next, the sediment transport function (under parameters) should be chosen based upon hydraulic conditions of the reach. I have used Engelund-Hansen, Laursen and Yang’s sand and gravel with reasonable results. Since Yang’s equation was the best method for my hydraulic geometry (fine sand to coarse gravel bed), I used this.
4. Still under sediment control, I used a single iteration for each sediment calculation. Each time I used more than one, the model became unstable more quickly. Balance diffusion was also check, with bed convergence parameters set to 0.1 and 0.1. More stringent convergence criteria may be possible, but for my purpose, this was adequate.
5. Under the general tab in model control, semi coupled simulation was used with the initial conditions file being the latest hydrodynamic solution produced by the steering module. This is important, because for each successive run under semi coupled, the initial conditions will be updated and allowed to achieve final convergence. Also, set the run to dynamic, not steady state.
6. Under parameter times, I used 15 iterations for the hydrodymamics, with a 6.0 hour run at a time step on 0.1 hours. For sediment transport larger time steps are necessary to show bed evolution. Time steps are not important since the solution method is implicit meaning the Courant time step is not applicable. The time integration factor was left at the default.
7. Exit the model control parameters and check the material properties under the FESWMS menu. Manning’s values are arbitrary and work for all simulations. However, under turbulence parameters, artificially high values are necessary to ensure the hydrodynamics are as stable as possible.
8. For the upstream boundary condition use a constant flow with sediment options set to equilibrium conditions applied and initial bed elevation fixed. I tried clear water and it worked, but more realistic results occurred with equilibrium conditions. For the downstream boundary condition, equilibrium rates are also applied.
9.The model is now ready to run. Go to FESWMS menu and select run simulation. I always received the bad aspect triangles, but I ignored it and ran the model. Wait a bit and the model will hit an error (for me it was around 0.5 hours). This is not a problem, since the *.FLO file as the initial conditions will now have 5 times steps in it. Just rerun the model and it should progress further. Mine reached approximately 2.0 hours the second time, 3.5 hours the third, etc, until it finally ran to completion. With a complete *.SDI file, use this as the initial sediment condition if the sediment transport function is changed.
10. It is also recommended for maximum stability that you have zero dry nodes at all time steps. This is very important especially at the boundaries. Every time I had even one dry node, the sediment transport run became unstable.
I hope this helps for anyone else who was as frustrated as I was.