Timestepping¶

This input file section sets the start time, end time, and time step. The time step can be altered in user coding, see below. Changing the time step increment will cause the next time step to use a first order time integration method (Euler) to avoid consistency errors in solvers which support this (e.g. the IPCS-A solver).

time:
    dt: 0.01
    tmax: 60.0

dt: Time step

tstart: Starting time, default 0.0

tmax: Maximum simulation time, simulation will stop when t > tmax

Adaptive time stepping¶

Adaptive time stepping can be implemented by use of hooks in the input file, see Hooks - run custom code at given times during a simulation for details.

A very simple example:

hooks:
    pre_timestep:
    -   name: decrease time step
        code: |
            if t > 10:
                simulation.input.set_value('time/dt', 0.005)

An example of adapting the time step depending on the Courant number:

hooks:
    post_timestep:
    -   name: Adaptive time step
        code: |
            # Configuration
            TARGET_CO = 0.3
            MINIMUM_DT = 0.0001

            # Get the cell based Courant number for all previous time steps
            all_Co = simulation.reporting.timestep_xy_reports.get('Co', [0])

            # Courant number for the last time step
            Co = all_Co[-1]

            # Maximum Courant number in the last 10 time steps
            Co_window = numpy.max(all_Co[-10:])

            # Compute new dt based on the Courant numbers
            new_dt = None
            if Co > TARGET_CO:
                new_dt = dt / 2
            elif Co_window < TARGET_CO / 6:
                new_dt = dt * 2

            # Change the dt if required (and the new dt is not below minimum)
            if new_dt is not None and new_dt >= MINIMUM_DT:
                simulation.log.info('Changing dt from %.5f to %.5f' % (dt, new_dt))
                simulation.input.set_value('time/dt', new_dt)