Navigation:  Numerical Modeling Workflow - Finite Difference Grids > Define Boundary Conditions >

Boundary Conditions - Theory

Print this Topic Previous page Return to chapter overview Next page


A discussion on how boundary conditions are conceptualized and parameterized in Visual MODFLOW Flex can be found here .

Steady-State vs. Transient Flow Boundary Conditions

For transient simulations, MODFLOW requires the time element of the boundary conditions to be defined using Stress Period “counters” as opposed to using “real” times. As a result, each time interval for a transient model must be determined in terms of Stress Periods before any boundary condition data is defined. Unfortunately, accommodating this format is quite tedious because the data collected for parameters such as rainfall and groundwater recharge don’t always follow the same time schedules as data collected for other boundary conditions, such as well pumping rates and surface water levels. This approach also makes it difficult to utilize raw field data collected and recorded in terms of real times.

In Visual MODFLOW Flex, a Time Period is similar to a Stress Period, but with two important exceptions:

1.A Time Period is defined using real times and real time units, and

2.Each boundary condition grid cell may contain different Time Periods


The advantage of this approach is the ability to clearly see the magnitude of time for each Time Period (as opposed to interpreting data such as “from Stress Period 1 to Stress Period 2”), and it facilitates more convenient methods for importing raw data from different boundary condition types.

Each group of boundary condition grid cells requires a minimum of one Time Period of data containing a Start Time, a Stop Time, and a complete set of data for the selected boundary condition type (the required data for each boundary condition type are described later in this section). For steady-state simulations, Visual MODFLOW Flex requires data for only a single Time Period, while for transient simulations, Visual MODFLOW Flex can accommodate an unlimited number of Time Periods.

For steady-state simulations, the Stop Time value is irrelevant because the term “steady-state” indicates that the model results are not changing with time. Therefore, a Stop Time value of 1 time unit is commonly used. However, if the model is going to be used to evaluate a transient simulation in the future, it is probably a better idea to give it a more realistic value corresponding to the potential time frame of interest.


If a steady-state simulation is run using a model containing transient boundary condition data, only the data from the first Time Period of each grid cell will be used for the steady-state conditions.  Furthermore, if the boundary condition consists of starting and ending values (e.g. starting head and ending head for constant head boundaries), the starting value will be used.


Incomplete Boundary Condition Schedules

If the defined period for a boundary condition in a transient model is of a shorter duration that the model itself, Visual MODFLOW Flex will assume that the boundary condition does not exist for the remaining duration. For example, a transient model is run for 10 years, and the recharge boundary condition is defined only for a period up to 7 years, Visual MODFLOW Flex will assume this boundary condition does not exist (i.e. recharge = 0) for the remaining 3 years of the simulation.


The exceptions to this rule are the Constant Head and Constant Concentration Boundaries, which must be defined for the entire simulation.



Note: For a steady-state simulation, a minimum of one active grid cell in the model MUST contain a head-dependent boundary condition type. Otherwise, the model is indeterminate and the solution will not converge. This head-dependent boundary condition acts as a reference head for all calculations. The head-dependent boundary condition type can be one of the following:


Constant Head (CHD)

River (RIV)

Stream (STR)

General-Head (GHB)

Lake (LAK)


For a transient simulation, the specified initial heads are sufficient for a determinant solution.


Page url: