16.10.2. Using Interrupt Control in Cases with Transient Convergence Behavior

Some steady-state or transient cases will not converge to the target residual levels recommended in Residual Type and Target Levels. In some cases, the lack of convergence can be attributed to inherent unsteadiness in the solution. Such cases are considered to exhibit transient convergence behavior.

If you expect your case to exhibit transient convergence behavior, especially periodic behavior, you should consider setting interrupt control conditions. For such cases, you can stop the run based on trends in monitored quantities. For example, you could make an interrupt control condition that stops a run when all of the following criteria are met:

  • At least 100 iterations have passed.

  • The arithmetic averages of a monitor of efficiency over two intervals of different size, 25 and 50 iterations, are within a certain relative tolerance, 1.0e-04.

  • The coefficient of variation (the ratio of the standard deviation to the mean) of a monitor of efficiency over an interval of 50 iterations is below a certain value, 1.0e-04.


Note:  An interrupt control condition cannot be computed and has no effect until all of the monitor statistics that are involved in its definition can be computed. If you set a minimum number of iterations (100 in this case), then that number should be at least the size of the interval required for all statistics to be computed (50 iterations in this case).


In order to set up this interrupt control condition in CFX-Pre, you could follow this procedure:

  1. Create an expression for efficiency named Eff.

    For details, see Creating an Expression in the CFX-Pre User's Guide.

  2. Create a monitor named MP Eff 25.

    For details, see Setting up Monitors in the CFX-Pre User's Guide.

  3. Configure the following setting(s):

    Setting

    Value

    MP Eff 25 > Option

    Expression

    MP Eff 25 > Expression Value

    Eff

    MP Eff 25 > Monitor Statistics

    (Selected) [ a ]

    MP Eff 25 > Monitor Statistics > Statistics List

    Arithmetic Average

    MP Eff 25 > Monitor Statistics > Interval Definition > Number of Iterations

    25

    1. Monitor statistics enable you to evaluate statistical quantities for an expression value over a moving interval. For details, see [Monitor Name]: Monitor Statistics in the CFX-Pre User's Guide.

  4. Create a monitor named MP Eff 50.

  5. Configure the following setting(s):

    Setting

    Value

    MP Eff 50 > Option

    Expression

    MP Eff 50 > Expression Value

    Eff

    MP Eff 50 > Monitor Statistics

    (Selected)

    MP Eff 50 > Monitor Statistics > Statistics List

    Arithmetic Average, Standard Deviation

    MP Eff 50 > Monitor Statistics > Interval Definition > Number of Iterations

    50

  6. Create an expression named Avg Eff 25 with the following definition:

    probe(Expression Value.Arithmetic Average)@MP Eff 25

    This expression calculates the arithmetic average for MP Eff 25 over the current interval (25 iterations). For details, see Using Monitors in CEL in the CFX-Pre User's Guide.

  7. Create an expression named Avg Eff 50 with the following definition:

    probe(Expression Value.Arithmetic Average)@MP Eff 50
  8. Create an expression named StdDev Eff 50 with the following definition:

    probe(Expression Value.Standard Deviation)@MP Eff 50
  9. Create the interrupt control condition by accepting the default name and using the following logical expression:

    aitern>100 && abs((Avg Eff 25-Avg Eff 50)/Avg Eff 25)<0.0001 && abs(StdDev Eff 50/Avg Eff 50)<0.0001

    The interrupt control settings are described in Interrupt Control in the CFX-Pre User's Guide.