Maximum K evaluations

The largest number of evaluations of $K$ for a perforated tube with the grazing correction, including the one at the end of the upwind start-up phase. After the last one (or after the first check that moves $K$ by less than 2 %) the run continues to convergence with the $K$ it has.

Last updated 2026-11-03

Theory

Evaluation 1 is made at the end of the upwind phase and costs no extra restart. Evaluations 2 ... $n_{max}$ are checks during the final phase, one every K check interval iterations; each restarts the solver only if $K$ moved by more than 2 %. With several perforated tubes one schedule applies to all of them: the largest $n_{max}$ among them.

The pressure-drop window of the convergence rule starts the settle iterations after the LAST real restart, so every restart delays the earliest possible convergence. The list of evaluations (iteration, target and relaxed $K$ , and whether a restart was made) is kept in logs/perforate_k.json (_rounds).

Equations

n_{checks} \le n_{max} - 1, \qquad i_{window} = i_{restart,last} + n_{settle}

(1)

where	meaning	unit
$n_{max}$	Maximum number of evaluations (this setting)	-
$n_{checks}$	Checks made during the final phase	-
$i_{window}$	First iteration the convergence window may use	-
$i_{restart,last}$	Iteration of the last restart (the scheme switch when $K$ never restarted)	-
$n_{settle}$	Settle iterations after a restart	iterations

Default and rule

Default: 5
Rule: 5.
Source: A run-control rule of the program. On the two-baffle test muffler the first check already moved $K$ by only 0.09 %, so the checks ended after two of the five evaluations. The number is a cap, not a tuned value.

Valid range 1-50

Values

1-50. With 1, only the evaluation at the end of the upwind phase is made.

Continued runs

A run continued with Continue run (resume) performs its final phase without $K$ checks.

Theory

Equations

Default and rule

Valid range 1-50

Related 3