*CONTROLS

Keyword type: step

This option is used to change the iteration control parameters. It should only be used by those users who know what they are doing and are expert in the field. A detailed description of the convergence criteria is given in Section 6.10. There are two, mutually exclusive parameter: PARAMETERS and RESET. The RESET parameter resets the control parameters to their defaults. The parameter PARAMETERS is used to change the defaults. It can take the value TIME INCREMENTATION, FIELD, LINE SEARCH, NETWORK, CFD or CONTACT. If the TIME INCREMENTATION value is selected, the number of iterations before certain actions are taken (e.g. the number of divergent iterations before the increment is reattempted) can be changed and effect of these actions (e.g. the increment size is divided by two). The FIELD parameter can be used to change the convergence criteria themselves.

LINE SEARCH can be used to change the line search parameters (only for face-to-face penalty contact). The line search parameter scales the correction to the solution calculated by the Newton-Raphson algorithm such that the residual force is orthogonal to the correction. This requires the solution of a nonlinear equation, and consequently an iterative procedure. In CalculiX this procedure is approximated by a linear connection between:

• the scalar product of the residual force from the last iteration with the solution correction in the present iteration (corresponds to a line search parameter of zero) and
• the scalar product of the residual force in the present iteration with the solution correction in the present iteration (corresponds to a line search parameter of one).

For details of the line seach algorithm the reader is referred to [84].

With the NETWORK parameter the convergence criteria for network iterations can be changed. The parameters $c_{1t}$, $c_{1f}$ and $c_{1p}$ express the fraction of the mean energy balance, mass balance and element balance terms the energy balance residual, the mass balance residual and the element balance residual is not allowed to exceed, respectively. The parameters $c_{2t}$, $c_{2f}$, $c_{2p}$ and $c_{2a}$ is the fraction of the change in temperature, mass flow, pressure and geometry since the beginning of the increment the temperature, mass flow, pressure and geometry change in the actual network iteration is not allowed to exceed, respectively.

With the CFD parameter the maximum number of iterations in certain fluid loops can be influenced. A fluid calculation within CalculiX is triggered at the start of a new mechanical increment. This increment is subdivided into fluid increments based on the physical fluid properties. For each fluid increment iterations are performed. In steady state calculations only one iteration is performed, since no time-accurate solution is needed. In transient calculations iterations are performed until convergence of the fluid increment or until the maximum allowed number of iterations is reached. This is the first parameter $iitt$ (``transient''). In fluid calculations the unknowns in the equation systems are the quantities (velocity..) at the element centers. The values at the face centers and the gradients are calculated based on these element center quantities. In case the mesh is not orthogonal, iterations have to be performed. The number of these iterations is expressed by $iitg$ (``geometry'') and $iitp$ (taking non-orthogonality into account in the pressure correction equation, ``pressure''). This is the second and third parameter. For a perfectly rectangular grid these values can be set to zero.

Finally, the CONTACT parameter is used to change defaults in the face-to-face penalty contact convergence algorithm (cf. Section 6.10.2). This relates to

• the maximum relative difference in number of contact elements to allow for convergence (delcon). The corresponding absolute difference, which may not be exceeded is defined as the number of contact elements in the previous iteration times delcon.
• the fraction of contact elements which is removed in an aleatoric way before repeting an increment in case of a local mimimum in the solution (alea)
• the integer factor by which the normal spring stiffness (in case of linear pressure-overclosure) and stick slope are reduced in case of divergence or too slow convergence (kscalemax)
• the maximum number of iterions per increment (itf2f).

First line:

• *CONTROLS
• Enter the PARAMETERS parameter and its value, or the RESET parameter.

There are no subsequent lines if the parameter RESET is selected.

Following lines if PARAMETERS=TIME INCREMENTATION is selected:

Second line:

• $I_0$ iteration after which a check is made whether the residuals increase in two consecutive iterations (default: 4). If so, the increment is reattempted with $D_f$ times its size.
• $I_R$ iteration after which a logarithmic convergence check is performed in each iteration (default: 8). If more than $I_C$ iterations are needed, the increment is reattempted with $D_C$ its size.
• $I_P$ iteration after which the residual tolerance $R^\alpha_p$ is used instead of $R^\alpha_n$ (default: 9).
• $I_C$ maximum number of iterations allowed (default: 16).
• $I_L$ number of iterations after which the size of the subsequent increment will be reduced (default: 10).
• $I_G$ maximum number of iterations allowed in two consecutive increments for the size of the next increment to be increased (default: 4).
• $I_S$ Currently not used.
• $I_A$ Maximum number of cutbacks per increment (default: 5). A cutback is a reattempted increment.
• $I_J$ Currently not used.
• $I_T$ Currently not used.

Third line:

• $D_f$ Cutback factor if the solution seems to diverge(default: 0.25).
• $D_C$ Cutback factor if the logarithmic extrapolation predicts too many iterations (default: 0.5).
• $D_B$ Cutback factor for the next increment if more than $I_L$ iterations were needed in the current increment (default: 0.75).
• $D_A$ Cutback factor if the temperature change in two subsequent increments exceeds DELTMX (default: 0.85).
• $D_S$ Currently not used.
• $D_H$ Currently not used.
• $D_D$ Factor by which the next increment will be increased if less than $I_G$ iterations are needed in two consecutive increments (default: 1.5).
• $W_G$ Currently not used.

Following line if PARAMETERS=FIELD is selected:

Second line:

• $R^{\alpha}_n$ Convergence criterion for the ratio of the largest residual to the average force (default: 0.005). The average force is defined as the average over all increments in the present step of the instantaneous force. The instantaneous force in an increment is defined as the mean of the absolute value of the nodal force components within all elements.
• $C^{\alpha}_n$ Convergence criterion for the ratio of the largest solution correction to the largest incremental solution value (default: 0.01).
• $q^{\alpha}_0$ Initial value at the start of a new step of the time average force (default: the time average force from the previous steps or 0.01 for the first step).
• $q^{\alpha}_u$ user-defined average force. If defined, the calculation of the average force is replaced by this value.
• $R^{\alpha}_p$ Alternative residual convergence criterion to be used after $I_P$ iterations instead of $R^\alpha_n$ (default: 0.02).
• $\epsilon^{\alpha}$ Criterion for zero flux relative to $q^{\alpha}$ (default: $10^{-5}$).
• $C^{\alpha}_{\epsilon}$ Convergence criterion for the ratio of the largest solution correction to the largest incremental solution value in case of zero flux (default: $10^{-3}$).
• $R^{\alpha}_l$ Convergence criterion for the ratio of the largest residual to the average force for convergence in a single iteration (default: $10^{-8}$).

Following line if PARAMETERS=LINE SEARCH is selected:

Second line:

• not used.
• $s^{ls}_{max}$ Maximum value of the line search parameter (default: 1.01).
• $s^{ls}_{min}$ Minimum value of the line search parameter (default: 0.25).
• not used.
• not used.

Following line if PARAMETERS=NETWORK is selected:

Second line:

• $c_{1t}$ (default: $5 \cdot 10^{-7}$).
• $c_{1f}$ (default: $5 \cdot 10^{-7}$).
• $c_{1p}$ (default: $5 \cdot 10^{-7}$).
• $c_{2t}$ (default: $5 \cdot 10^{-7}$).
• $c_{2f}$ (default: $5 \cdot 10^{-7}$).
• $c_{2p}$ (default: $5 \cdot 10^{-7}$).
• $c_{2a}$ (default: $5 \cdot 10^{-7}$).

Third line:

• $a_{1t}$ (default: $10^{20}$).
• $a_{1f}$ (default: $10^{20}$).
• $a_{1p}$ (default: $10^{20}$).
• $a_{2t}$ (default: $10^{20}$).
• $a_{2f}$ (default: $10^{20}$).
• $a_{2p}$ (default: $10^{20}$).
• $a_{2ac}$ (default: $10^{20}$).

Following line if PARAMETERS=CFD is selected:

Second line:

• $iitt$ (default: $20$).
• $iitg$ (default: $1$).
• $iitp$ (default: $1$).

Following line if PARAMETERS=CONTACT is selected:

Second line:

• delcon ($\ge 0$; default: $0.001$).
• alea ($0 \le$ alea $\le 1$; default: $0.1$).
• kscalemax ($\ge 1$, integer; default: $100$).
• itf2f ($\ge 1$, integer; default: $60$).

Example:

*CONTROLS,PARAMETERS=FIELD
1.e30,1.e30,0.01,,0.02,1.e-5,1.e-3,1.e-8

leads to convergence in just one iteration since nearly any residuals are accepted for convergence ( $R^{\alpha}_n=10^{30}$ and $C^{\alpha}_n=10^{30}$.

Example files: beammrco.