User Forum of Software BASEMENT

ZoeUniBase

User

Registered: Yesterday

Posts: 1

Sediment level exceeds dike level in v4.2 but works v2.8.2 (same n, α)

Hello,

I am running a 1D morphodynamic calibration study (BASEMD, Meyer-Peter & Müller bedload formula) over a grid of Manning's n and bedload factor alpha values, on a laboratory flume geometry (99 cross-sections, dx = 5 cm, slope = 13.7‰).

For certain (n, alpha) combinations, BASEMENT v4.2.0 fails with the following error:

ERROR -> Sediment level exceeds the lower dike level in cross section: s9
   (dikemax: 111.512, z: 1.97883e+09)
   The program cannot continue.

The simulation stops shortly after the time loop starts (around t = 2 s simulated time), with the bed elevation diverging to a clearly non-physical value (~2e9 m).

What is confusing is that I ran the same model.json configuration using BASEMENT v2.8.2 (with the same geometry, boundary conditions, n and alpha values), and it completed successfully without any errors or instability, despite the same parameter combinations failing in v4.2.0. For example, for n = 0.019 and alpha = 2.

Setup summary:
- Module: BASEMD, BASECHAIN_1D
- Grid search (min, max, step) :
    - Manning friction (n) : (0.015, 0.03, 0.0005),
    - Bedload formula: mpm, bedload_factor (alpha) : (1,   3,   0.05),
- Upstream boundary: sediment_discharge (Sedigraph.txt) and hydrograph (Hydrograph.txt)
- Downstream boundary: IODown and hqrelation
- Initial condition: continue from a clearwater restart file
- CFL = 0.8, upwind = 0.5

Has anything changed in the BASEMD bedload transport implementation, the theta_critical initialization, or the timestep adaptation between v2.8.2 and v4.2.0 that could explain this difference in stability for identical input parameters?

I am happy to provide a minimal model.json/simulation.json example or a screenshot of the relevant geometry if that would help diagnose this.

Thank you for your help.

Best regards,
Zoé Ury