SZ Computational Control Parameters         

For the saturated zone in MIKE SHE there are two solvers to choose from:

·         the pre-conditioned conjugate gradient method, and

·         the successive over-relaxation method.

The Successive Over-relaxation solver is the original solver in MIKE SHE and the Pre-conditioned Conjugate Gradient Solver is based on the USGS’s PCG solver for MODFLOW (Hill, 1990).

Steady-state vs Transient Simulations

The Solver type controls whether or not the simulation is run as a Steady-state model or not - if you chose the Pre-conditioned Conjugate Gradient-Steady-State option then the simulation will be run in steady-state. Otherwise, the simulation will be run as a transient simulation.

If the SZ simulation is steady-state, then the PCG solver is the only solver available. Although the same options are available for both the steady-state and the transient PCG solvers the optimal parameters or combination of parameters and options is most likely different in the two cases. Thus, the recommended settings are different in both cases.

Iteration Control

The iteration procedure can be stopped when either the iteration stop criteria are reached or when the maximum number of iterations is reached. The itera­tion stop criteria consist of a mass balance criteria and a head criteria. Both of these criteria must be chosen carefully to ensure that the solution has con­verged to the correct solution.

The default option settings normally perform well in most applications. Usu­ally there is no need for changes. Changes to the default options should not be done unless the solution does not converge or convergence is extremely slow.

Maximum number of iterations - The maximum number of iterations should be sufficiently large to avoid water balance errors due to non-conver­gence.

Maximum head change per iteration - The head criteria determines the accuracy of the solution. The computational time is very dependent on the value used. A value of 0.01m (0.025ft) is usually sufficient. During the initial model calibration a higher stop criteria can be used. The sen­sitivity of the head stop criteria should always be examined.

Maximum residual error - The maximum residual error is the tolerable mass balance error, which should be low but sufficiently high that the number of iterations is not excessive. A value of 0.001m/d is usually good for regional groundwater studies. In smaller scale applications, where sol­ute transport will be investigated the mass balance criteria should be reduced, for example, to 0.0001 or 0.00001m/d. In general, a larger mass balance criteria should be used during model calibration to keep the initial simulation times shorter. For scenario calculations, the mass balance criteria can be reduced to ensure more accurate simulations and smaller mass balance errors. The SZ water balance should always be checked at the end of the simulation to ensure that the mass bal­ance criteria used was reasonable.

Sink de-activation in drying cells

Saturated thickness threshold - To avoid numerical stability problems the minimum depth of water in a cell should always be greater than zero. However, if the water depth is close to zero, then sinks in the cell, such as wells, should be turned off, since the cell is effectively ‘dry’. This value is the minimum depth of water in the cell and the depth at which the sinks are deactivated.

Maximum exchange from river during one time step

Max. fraction of H-point volume - If you are simulating rivers with MIKE Hydro River, then this represents the maximum water that can be removed from the river during one SZ time step. Removing larger amounts of water could effectively dry out the river. If this occurs, then the SZ solver will issue a warning and only this fraction of water will be removed, which prevents rapid drying out of the river during a single time step.