Simulation

Introduction

The FEFLOW user interface keeps all visualization options available during the simulation. The simulation progress can thus be conveniently monitored and problems can be detected early.

FEFLOW does not separately store initial conditions of hydraulic head or concentration/age/temperature during the simulation. So the hydraulic-head process variable contains the initial head values before the simulation, the current hydraulic head results at each time step of a transient simulation, and the final hydraulic head after the simulation. These final results are also retained when leaving the simulation mode by clicking Stop in the Simula­tion toolbar.

 

If a given simulation record does not contain all the time steps, postprocess­ing methods that cover multiple time steps, such as transient pathline analy­sis or time-integrated budgets, can only give approximate answers.

Model Check

Before running the model, all input parameters and other model properties should be thoroughly checked. A basic overview of the model characteristics is provided in the Problem Summary page in the Problem Settings dialog. The icon preceding each material property in the Data panel indicates whether the respective parameter is homogeneous or heterogeneous and time constant or time dependent.

Results Output

Simulation results can be stored in two different formats: a reduced results file (*.dar) and a full simulation record (*.dac). Both results files can be recorded at the same time.

The reduced format contains text output for the observation points and well locations only. It is especially useful for automatic output processing with user-programmed scripts or optimization software such as PEST. Reduced results files always contain values for all the time steps.

In contrast, the full simulation record contains all nodal values of the primary variables (hydraulic head, concentration, age, temperature) and of the flow velocities. By default, results at all time steps are stored. It is possible, how­ever, to skip a number of time steps between each output, or to specify the simulation moments for which output is desired. In the latter case, the auto­matic time-stepping scheme ensures that results are calculated exactly at the prescribed output moments.

It is also possible to record a results file that only contains the final time step.

Running the Simulation

The handling of the interface is as easy as starting a tape or video recorder: The Start button in the Simulation toolbar starts the simulation, the Pause button pauses it, the Stop button exits the simulation mode, and if the Record button has been pressed before starting the simulation, the results are recorded in a file while the simulation runs.

All visualization and also some results-evaluation features ( Rate- and Period-Budget panels, Subdomain Boundary Rate and Subdomain Boundary Period panels, Content panel) can be used during the simulation. It should be considered, however, that some of these features might slow down the simulation. Especially more elaborate visualization options such as isosurfaces or budgeting may require significant calculation effort.

After the simulation finishes, FEFLOW remains in paused simulator mode. Thus simulation results such as flow velocities and the solution of the last time step remain available for postprocessing. Only when exiting the simula­tion mode via Stop these data are removed from memory, and stream­line/pathline calculation or budget analysis are no longer possible. Any postprocessing then requires loading of the full simulation-record file (*.dac).

 Convergence

When solving nonlinear problems for steady state, or for transient conditions with constant or varying time steps, an Error-Norm chart is shown. Once the change in results between two subsequent iterations is below a certain error tolerance, the simulation is terminated. FEFLOW shows a message in the Log panel if no convergence was achieved within the defined maximum number of iterations.

In transient simulations with an automatic time-stepping procedure, the time step is shortened if necessary to achieve convergence. Only if repeated time-step reductions do not improve the situation, a log message is shown indicat­ing that time-step reduction failed. The simulation is continued with the results of the last calculation for the time step. In general, models that show conver­gence problems even after greatly reducing the time-step length should be checked for errors in the input data or for insufficient mesh resolution.